Project Final Report

PROJECT FINAL REPORT

Grant Agreement number: 317532

Project acronym: MULTIPLEX Project title: Foundational Research on MULTIlevel comPLEX networks and systems Funding Scheme: FP7-ICT-2011-8 Collaborative Project Period covered: 48 months from 01/11/2012 to 31/10/2016 Name of the scientific representative of the project's co-ordinator1, Prof. Guido Caldarelli Title and Organisation: NATIONAL RESEARCH COUNCIL OF ITALY– INSTITUTE OF COMPLEX SYSTEM (CNR- ISC) Tel: +390649913120 Fax:+39-064463158 E-mail: [email protected]

Project websiteError! Bookmark not defined. address: http://www.multiproject.eu

1 Usually the contact person of the coordinator as specified in Art. 8.1. of the Grant Agreement. 4.1 Final publishable summary report

EXECUTIVE SUMMARY

MULTIPLEX project lasted 4 years and gathered a Consortium composed by 23 different nodes, resulting in a project both longer and larger with respect to the “usual” FET scheme. The figures of the projects are impressive: • 23 different nodes in 16 different European Countries • It produced more than 400 scientific publications in international Journals, including high end journals like Physical Review (Letters, X, E), Nature, Nature Communication, PNAS, Science Advances, Nature Scientific Reports, PLOS. • 4 books have been published by participants plus the final one in preparation • Results have been disseminated by means of Schools, Conferences, Workshops and Sessions organised within the largest and most important events in the community: amongst which APS March meeting, NETSCI Conference, Conference on Complex Systems (CCS). Furthermore, two conferences, Lucca (E)CCS’14, and Zaragoza’s NETSCI’15 have been organised by the participants of Consortium (G. Caldarelli and Y. Moreno) • MULTIPLEX had a significative impact on regulatory bodies of the communities. The complex System Society is directed by the President (Y. Moreno) and by two vice presidents (A. Barrat, G. Caldarelli) all key players in the Consortium. G. Caldarelli is also member of the board of the Network Society and in the Statistical and Nonlinear Physics Division of the European Physical Society (SNPD-EPS) • About 30 researchers have been trained in a period fundamental for their career. Regardless the numbers listed above, the main scientific results is the relevant knowledge advancement in the field of multilevel complex systems. With respect to the starting point, 4 years ago, we can now: • Describe mathematically multilevel and multi-layered networks; • Have a description of the way in which the giant percolation clusters appears; • Describe mathematically simple dynamical processes as disease diffusion, information spreading and random walks on these structures; • Have available a set of models able to describe both the topology and the dynamics described above; • Applications realised for particular cases studies as Twitter and Facebook • Characterisation and measurement of misinformation growth and development; • Quantification of multi-layer systems of interest in finance as interbank network and CDS structure of connections; • Realisation of a procedure of network reconstruction for a variety of physical and social systems. Concluding, MULTIPLEX significantly contributed to the advancement of knowledge, showing how cutting-edge problems of finance, social systems, and infrastructural networks can be easily treated by means of the multilevel complex systems framework. 21

SUMMARY DESCRIPTION OF PROJECT AND OBJECTIVES

MULTIPLEX PROJECT (http://www.multiplexproject.eu) Prof.M Guido Caldarelli [email protected] +39-0649913120

Many real-world complex systems consist of multi-level networks because they can be conceived as interacting networks. For instance, in infrastructure systems the power grid and the internet are highly interdependent; in financial systems, the credit market, the derivative market and the real economy can reverberate shocks from one to the other; in social systems, rumours and political ideas spread back and forth between the on-line social networks and the off-line social networks. The MULTIPLEX project aims to provide the foundation to the mathematical and algorithmic theory of certain classes of multi-level complex systems. In particular, we focus on those systems that can be described by networks or networks of networks. The discipline of Complex Networks emerged in the last 15 years as a new and truly interdisciplinary field. One of the reasons of the growing scientific interest in this field, is the promising tools that can help us making some sense out of the data deluge in the modern society, as well as helping us in understanding the opportunities and the risks emerging from the growing interdependence of our technical, social, and economic systems. Yet we do not have a coherent theory of multi-level complex systems. The MULTIPLEX Consortium brings together most of the world leading scientists active in the community of Complex Networks, including physicists, computer scientists, economists, and mathematicians. Aware of the challenges involved, we believe to be in a good position to try and make a significant advance in the theory of Multi-level Complex Systems.

Theory of Socio-Economic Networks

UNIL AALTO IMT ETHZ Theory of Complex Networks BIU ISI UNIVIE UA CEU MIW UW

URV Theory of ICT Networks UZ UPB CNRS CTI AUTH LIMS UNIROMA1

We believe that this approach is general enough to describe most of the situations of interest. Nowadays, systems that are easily described by complex networks span across the digital and physical domains, evolve at a fast rate, create new markets, social systems and professional communities. In most of the cases, all these structures interact and affect each other. For instance, Internet cannot be studied in isolation from WWW or Twitter and Facebook. In this context, this project will investigate the onset and evolution of long-range correlation properties in real and artificial systems. This means defining a restricted set of topological/dynamical properties that a complex system (modelled as a network) must have and analyse which basic ingredients determine their formation. To reach this goal we prepared a theoretical path that starting from the basic requirements for investigation, moves to more refined questions. Naturally, we aim at predicting the behaviour of these systems. This feature will be obtained once suitable models have been proposed and validated. Typically this process involves a feedback between theoretical insights, data analysis and numerical simulations, and/or analytical computations. Also in this case only very preliminary results have been obtained and the aim of this proposal is to fill this gap. Another important aspect of our proposal is to investigate the computational and algorithmic principles underlying multi-level complex networks in the ICT domain. The topological properties of the network components have been extensively investigated in the recent past and several evolving models reproducing the observed degree, cliquishness and clustering coefficient have been proposed. However, ICT mediated complex networks display a larger number of specific properties that should carefully be considered in our study. First of all these networks are operated and stored by computer algorithms having only a local view of the networks. It follows that any realistic effort to describe such multi-level networks should lead to systems that can be efficiently constructed and explored by local distributed algorithms. A related question is whether these complex multi-level networks allow a fast spreading of information and epidemic phenomena and whether we observe fast convergence of the dynamics induced from the forces that drive the evolution of the multi-level network. In mathematical terms, the signature of complexity in networks is the appearance of regularities at multiple scales; e.g. spatial and temporal correlations between topological quantities, such as the nodes’ degree. For example, in disease spreading phenomena (e.g., the flu), the hubs of the contact networks between individuals in a population play a preponderant role in the various waves of the disease spreading. At a higher correlation level, two-points degree correlations determine the topological mixing assortativity properties of the network, which may slow down or enhance propagation phenomena. Moreover, in a general dynamical perspective, a process taking place on the network might co-evolve with the network itself. A feedback loop can take place between the structure of the network and what happens on it (e.g., the disappearance of edges/nodes can slow down the propagation of diseases that in turn causes the deletion of other edges/nodes) [Garlaschelli et al. 2007, Gross et al. 2008]. Correlations among the state variables of the nodes during dynamical processes can also account for causally-connected domino effects, important both in epidemics and information spreading.

Based on these considerations, we define the following main objectives of our project: O1. Mathematical Description Of Networks Of Networks Develop mathematical and algorithmic frameworks that will enable us to describe, both analytically and numerically, multi-level complex networks. To this aim, appropriate variables and parameters will be identified in order to determine the topology and to describe the evolution of such systems, as well as to optimize some of their functionalities. The set of new quantities will allow to: measure the persistence over time of the system’s statistical properties; describe the properties of graphs in the infinite limit of their size; assess the coarse grain similarity across different systems and across different scales; detect multi-level communities; incorporate the notion of hypergraphs; account for multi-level cascading effects and multi-level rare events. Exempla: (1) Development of a scaling procedure to map different levels (e.g. from business activities at the level of firms to those in the world trade web); (2) Assessment of global properties from local properties (e.g. local certificates for scalable algorithms) Within objective O1, the extended project introduce null models of MLCN with specified constraints. These null models will be defined and generated as maximum-entropy ensembles of MLCN with given topological properties, and will enable us to assess which higher-order properties of a real-world MLCN can be traced back to simpler lower-level constraints. The introduction of multi-level null models is a key advance, not present in the original proposal. When extended to MLCN, this will lead to a rigorous information theory of multi-level complex systems. While studies of network failures usually analyse breakdowns of connected components in the network, we are going to study how breakdowns affect tasks and resources which are distributed over the network and which are collectively needed to fulfil a specific task. This is the program that is missing in complex network community even for the case of simple networks. We will introduce some new methods in order to assess the impact of network failures on such mutual functional relationships between the roles of different nodes. O2. Mathematical Description Of Controllability and Feedback Between Topology And Dynamics Multi-level networks evolve continuously as a result of feedback loops between endogenous dynamics internal to the system, exogenous forces shaping the overall structure, and the system’s topology. Our objective is: first to develop a mathematical framework to describe this behaviour; second to clarify to what extent the structure of a real (or computer generated) multi-level complex system can be reconstructed from the partial information about nodes, edges and dynamics. Once the key quantities describing the system are determined, we aim at assessing the conditions under which it is possible to control the future evolution of the system. Moreover, we will provide a rigorous characterisation in terms of limitations and potential for distributed algorithms, running on multi- level complex networks. Exempla: (1) Description of cascading phenomena depending on the connections among different levels in the system; (2) Algorithmic characterization of cascade size from local information. (3) Determine the space of events of multi-level complex networks; (4) Ensure controllability at various levels by means of scalable algorithms. Within objective O2, the extension will analyse the effects of different time scales on network robustness. As an extension to the original proposal, we will tackle the problem of network failure propagation on changing networks. O3 Modelling Building on the framework resulting from Objectives 1-2, develop models, which will be able to predict and optimize the propagation of information, opinions, influence, epidemics and socio- economic trends within multi-level complex networks. In particular, we will develop Game- Theoretical models of competitive agents and will analyse physical models that explain global behaviour from the interaction between neighbouring agents. Exempla: characterize speed and size of information spreading in terms of few relevant parameters that characterize the physical and combinatorial structure of multi-level complex system. Within objective O3, the extended project will introduce novel methods for modelling and construction of MLCN. We will develop text mining methods to discover new relations between nodes in complex networks from unstructured, textual data. We will use existing network data as background knowledge to analyse events in complex networks. In particular, we will develop data mining methods for descriptive and predictive analysis of rare events, such as banking crises. The network data, interconnectivity in particular, will be combined with textual data to generate rich feature vectors which are then used by machine learning algorithms to build descriptive and predictive models. From the textual data, we will analyse sentiment attributed to network nodes and study different models of sentiment propagation in MLCN.

O4. Validation on Real-World Data The theoretical framework developed in the project will help in identifying relevant quantities from large sets of data. We aim at progressing towards the definition of a general methodology that allows abstracting models of multi-level complex systems through analysis of large heterogeneous datasets. Conversely, the proposed models will be extensively validated on real-world datasets. Exempla: Abstract a multi-level model from data collected in networks of proximity contacts at the level of individuals, at the level of the cities and the level of connections in online social networks.

Within objective O4, the extension will analyse several new datasets that enhance existing real- world MLCN with big data originating from a continuous stream of unstructured news, blogs and tweets. The datasets encapsulates interaction between human behaviour and ICT at different time scales and will be used for validation. In particular, we will contribute TBytes of financial and economic news and blogs (timescale of days and weeks), tweets about stocks and exciting events (timescale of minutes and hours), historical data on banking crises enhanced with macroeconomic data (timescale of years), and multimedia repositories (VideoLectures - http://lis.irb.hr/challenge/, ScienceWISE - http://sciencewise.info). ScienceWISE was developed within the online article repository arXiv (http://arxiv.org) and is a semantic and dynamic network of scientific concepts linked by different contributing authors, who are in turn connected within each concept by co- authored articles. We will exploit the theory developed within the existing project to identify key structural changes aiming at the automated detection of scientific discoveries. Our approach will be tested against the data, using ‘objective’ external information, e.g. identifying publications leading to the award of Nobel Prizes.

DESCRIPTION

Work package 1 Foundational study of Structure & Topology of Multi-level Complex Systems Task T1.1 Foundational study of structure and topology 1.1.1 Analytic Results on Networks of Networks We have been engaged in a wide variety of research focusing on the foundational study and topology of interdependent networks, in particular in the influence of spatial embedding and clustering on the percolation properties of interdependent networks. This will lead to a set of n equations that will be solved analytically. We will generalize these n generating functions system to include other more realistic properties, such as inter-/intra-links correlations and clustering. We will evaluate the critical thresholds and the size of the giant cluster, as well as the fraction of the global giant cluster in each component. Along this path, regarding spatial embedding, recent studies show that in interdependent networks a very small failure in one network may lead to catastrophic consequences (as black-outs easily show). Above a critical fraction of interdependent nodes, even a single node failure can invoke cascading failures that may abruptly fragment the system, whereas below this critical dependency a failure of a few nodes leads only to a small amount of damage to the system. So far, research has focused on interdependent random networks without space limitations. 1.1.2 Improve robustness of coupled networks Research has focused on interdependent random networks without space limitations. However, many real systems, such as power grids and the Internet, are not random but spatially embedded. We analytically and numerically studied the stability of interdependent spatially embedded networks modeled as lattice networks 1.1.3 Aggregating multipartite networks: hypergraphs In 1990 Bender, Canfield and McKay gave an asymptotic formula for the number of connected labelled graphs with n vertices and m edges, whenever n and the nullity m−n+1 tend to infinity. Let Cr(n, t) be the number of connected r-uniform hypergraphs on [n] = {1, 2, . . . , n} with nullity t = (r − 1)m − n + 1, where m is the number of edges. For r > 3, asymptotic formulae for Cr(n, t) are known only for partial ranges of the parameters: Karónski and Luczak gave one when t = o(log n/ log log n), and Behrisch, Coja-Oghlan and Kang one for t = Θ(n). Here we prove such a formula for any r > 3 fixed, and any t = t(n) satisfying t = o(n) and t à∞ as n à∞, complementing the last result. This leaves open only the case t/n à∞, which we expect to be much simpler, and will consider in future work. r Our approach is probabilistic. Let r > 2 be fixed and let H n,p denote the random r-uniform hypergraph on [n] in which each possible edge is present independently with probability p. Let L1 and M1 be the numbers of r vertices and edges in the largest component of H n,p. We prove a local limit theorem giving an asymptotic formula for the probability that L1 and M1 take any given pair of values within the “typical” range, for any in the supercritical regime, i.e., when p=p(n)=(1+e)(r−2)!n−r+1 where e=e(n) satisfies e3n à∞ and eà0. This easily implies the counting result described above. Given two random hypergraphs, or two random tournaments of order n, how much (or little) can we make them overlap by placing them on the same vertex set? We give asymptotic answers to this question. Recently, we adapted exploration and martingale arguments of Nachmias and Peres, in turn based on ideas of Martin-Löf, Karp and Aldous, to prove asymptotic normality of the number L1 of vertices in the largest component L1 of the random r-uniform hypergraph throughout the supercritical regime. In this paper we take these arguments further to prove two new results: strong tail bounds on the distribution of L1, and joint asymptotic normality of L1 and the number M1 of edges of L1.

1.1.4 Structural transformation in the networks of networks Bootstrap percolation is a type of cellular automaton that has been used to model various physical phenomena, such as ferromagnetism. For each natural number r, the r-neighbour bootstrap process is an update rule for vertices of a graph in one of two states: ‘infected’ or ‘healthy’. In consecutive rounds, each healthy vertex with at least r infected neighbors becomes itself infected. Percolation is said to occur if every vertex is eventually infected. Usually, the starting set of infected vertices is chosen at random, with all vertices initially infected independently with probability p. In that case, given a graph G and infection threshold r, a quantity of interest is the critical probability, pc(G,r), at which percolation becomes likely to occur. We look at infinite trees and, answering a problem posed by Balogh, Peres and Pete, we show that for any b ≥ r and for any e > 0 there exists a tree T with branching number br(T) = b and critical probability pc(T, r) < e. However, this is false if we limit ourselves to the well-studied family of Galton–Watson trees. We show that for every r ≥ 2 there exists a constant cr > 0 such that if T is a Galton–Watson tree with branching number br(T) = b ≥ r then ! �! ! � � > � !!! ! � We also show that this bound is sharp up to a factor of O(b) by giving an explicit family of Galton–Watson ! ! trees with critical probability bounded from above by �!� !!! for some constant Cr > 0.

SUBCRITICAL U-BOOTSTRAP PERCOLATION MODELS HAVE NON-TRIVIAL PHASE TRANSITIONS Recently Bollobás, Smith and Uzzell introduced a broad class of percolation models called U-bootstrap percolation, which includes r-neighbour bootstrap percolation as a special case. They divided two-dimensional U-bootstrap percolation models into three classes – subcritical, critical and supercritical – and they proved that, like classical 2-neighbour bootstrap percolation, critical and supercritical U-bootstrap percolation models have trivial critical probabilities on ℤ!. They left open the question as to what happens in the case of subcritical families. In this paper we answer that question: we show that every subcritical U-bootstrap percolation model has a non-trivial critical probability on ℤ!. This is new except for a certain ‘degenerate’ subclass of symmetric models that can be coupled from below with oriented site percolation. Our results re- open the study of critical probabilities in bootstrap percolation on infinite lattices, and they allow one to ask many questions of subcritical bootstrap percolation models that are typically asked of site or bond percolation.

On the left, the equivalence of the update

family U1 and the DTBP mode: the dark grey site becomes infected when at least two of the light grey ones are.

On the right the stable set S1 For the

update family U1 (tick line) note that indeed every semicircle in S1 intersects

Int S1

UNIVERSALITY IN CELLULAR AUTOMATA ON A GRAPH An important and challenging problem in statistical physics, probability theory and combinatorics is to understand the typical global behavior of so-called ‘lattice models’, including cellular automata, percolation models, and spin models. Although these models are defined in terms of local interactions between the sites of the lattice, it is typically observed in simulations that, in fixed dimensions, the macroscopic behaviour of the models does not seem to depend on the precise nature of these local interactions. We study the class of monotone, two-state, deterministic cellular automata, in which sites are activated (or ‘infected’) by certain configurations of nearby infected sites. These models have close connections to statistical physics, and several specific examples have been extensively studied in recent years by both mathematicians and physicists. This general setting was studied only recently, however, by Bollobás, Smith and Uzzell, who showed that the family of all such ‘bootstrap percolation’ models on ℤ! can be naturally partitioned into three classes, which they termed subcritical, critical and supercritical. We determine the order of the threshold for percolation (complete occupation) for every critical bootstrap percolation model in two dimensions. This “universality” theorem includes as special cases results of Aizenman and Lebowitz, Gravner and Griffeath, Mountford, and van Enter and Hulshof, significantly strengthens bounds of Bollobás, Smith and Uzzell, and complements recent work of Balister, Bollobás, Przykucki and Smith on subcritical models. SELF-HEALING NETWORKS Self-healing networks are a model to describe distribution networks consisting of spanning trees with some backup links that can be activated in case of failure to ensure that most of the users will be still connected to the supplier. We show that in some region of the phase space determined by the probability of failure and that of activation of the backup links the giant component is ensured. We compute analytically the separation between the two behaviors of fragmented graphs and the giant component in the case of random graphs and square lattice. From an applicative point of view these results will help in designing a better distribution service and constitute a benchmark against which it is now possible to measure the optimization realized in real structures. We analyze the behavior of these trees by varying the two parameters defining them. This are the probability of failure of one edge on tree T and the rewiring probability for a backup link. We are able to compute under which condition of the parameters, the resulting graph have a giant component (i.e.a component of order W(n) as nà ∞). We consider mainly the cases when G is the complete graph, and when Gn is a square lattice. We present here some results (whose derivation is reported in the supplementary information attached) for a variety of trees, including the case when the tree is chosen uniformly at random from the set of all spanning trees. We consider the idea of selecting at random some “backup” links which we can activate so as to “repair” the network. The question is, how many backup links do we typically need. In the initial state, the source node (filled vertex, bottom left corner) is able to serve all nodes. The 4 dotted lines are backup edges that can be activated upon failure. The (red online) edge crossed by two short lines is disconnected with probability qT. (Central panel) failure of the edge disconnects all the nodes of a sub-tree. (Right Panel) By activating a single dormant backup edge (green) with probability pR, we recover connectivity for the whole system.

(Left Panel) the phase diagram for uniformly random trees on square lattice. (Central Panel) the phase diagram for uniformly random trees on Random Graphs. Simulated threshold functions for uniform random tree (black), and the left (red) and right (green) trees from possible best and worst cases for large pT shown in blue.

(LEFT) Conjectural best and worst trees for forming giant components when pT is small. (RIGHT) Conjectural best and worst trees for forming giant components when pT is larger.

Task T1.2 Classification of interdependent networks 1.2.1 Classification and characterization of real-world interdependent networks We employ the dk-series, a systematic basis of basic characteristics of network structure, to study the statistical dependencies between different network properties. The dk -series subsumes all the basic degree- based characteristics of networks (Figure 1). Indeed, the zero-th element of the dk -series, the 0k –“distribution", is just the average degree of the graph. The first element, the 1k -distribution, is the standard degree distribution, i.e. the number of nodes (subgraphs of size 1) of degree k in the network. The second element, the 2k-distribution, is the joint degree distribution, i. e. the number of subgraphs of size 2 (links) between nodes of degrees k1 and k2. The 2k -distribution thus defines 2-node degree correlations and network's assortativity. For arbitrary d the dk-distribution characterizes the degree k- correlations in d-sized subgraphs, thus including, on the one hand, the correlations of degrees of nodes located at hop distances below d and, on the other hand, the statistics of d-cliques. The special case d = 2.5 indicates that degree correlations are preserved up to second order and that, in addition, the function c(k), expressing the dependence of the average clustering coefficient of nodes of degree k on k, is also maintained. We consider six real networks – the Internet, US airport network, human protein interactions, technosocial web of trust, English word network, and an fMRI map of the human brain – and find that many important local and global structural properties of these networks are closely reproduced by dk -random graphs whose degree distributions, degree correlations, and clustering are as in the corresponding real network. In most cases, the considered networks are dk -random with d ≤ 2.5, i.e., d ≤ 2. 5 is enough to reproduce not only basic microscopic (local) properties but also mesoscopic and even macroscopic (global) network properties. 1.2.2 Multi-level networks with hidden metric spaces We analyze the hyperbolicity of real-world networks, a geometric quantity that measures if a space is negatively curved. We provide two improvements in our understanding of this quantity: first of all, in our interpretation, a hyperbolic network is “aristocratic”, since few elements “connect” the system, while a non- hyperbolic network has a more “democratic” structure with a larger number of crucial elements. The second contribution is the introduction of the average hyperbolicity of the neighbors of a given node. Through this definition, we outline an “influence area” for the vertices in the graph. We show that in real networks the influence area of the highest degree vertex is small in what we define “local” networks (i.e., social or peer-to- peer networks), and large in “global” networks (i.e., power grid, metabolic networks, or autonomous system networks). 1.2.3 Reconstruction and characterization of real-world interdependent networks Following the same lines as above, we approached the network reconstruction problem by proposing a general method that uses the knowledge of an intrinsic property of the nodes (indicated as fitness), and the number of connections of only a limited subset of nodes, in order to generate an ensemble of exponential random graphs that are representative of the real systems and that can be used to estimate its topological properties. We focused in particular on reconstructing the most basic properties that are commonly used to describe a network: density of links, assortativity, clustering. We tested the method on both benchmark synthetic networks and real economic and financial systems, finding a remarkable robustness with respect to the number of nodes used for calibration. The method thus represents a valuable tool for gaining insights on privacy- protected systems

Task T1.3 Multi-scale community structure 1.3.1 Multi-scale community detection: the hierarchy paradox We propose a method based on a compression of network flows that can identify modular flows both within and across layers in non-aggregated multilayer networks. Our numerical experiments on synthetic multilayer networks, with some layers originating from the same interaction process, show that the analysis fails for aggregated networks or when considering the layers separately, whereas the multilayer method accurately identifies modules across layers that originate from the same interaction process. We have proposed an extension of the classical Informap algorithm to multilayer networks. We introduced a method based on compression of network flows. The information-theoretic method generalizes the so-called map equation for networks with memory to take advantage of modular flows in multilayer networks. The framework generalizes straightforwardly because the information-theoretic machinery remains the same and only the non-Markovian flow model changes, with memory of the current layer rather than of the previous step. This approach therefore suggests a natural concept of communities in multilayer networks as groups of nodes that capture flows within and across layers for a relatively long time. The algorithm can be found integrated to our public available software at: http://muxviz.net/index.php

1.3.2 Modelling and evolution of multi2-scale community structure

(a) Grow / Shrink mmax pairs of nodes, the actual number of edges is drawn The model for generating networks with evolving community from a binomial distribution with parameters mmax and p. Then, we simply place this number of edges randomly to gen- structure we propose is based on the classic stochastic block model erate a sample from our ensemble. (SBM). It works as follows: a network is divided in a number q of The model implements the two fundamental classes of dynamic processes: growing/shrinking and merging/splittingof t=0 t=τ/4 t=τ/2 t=3τ/4 t=τ subgraphs and the nodes of the same subgraph are linked with a communities. By combining these two reversible types of processes one can capture the most common behaviors of dy- (b) Merge / Split probability pin, whereas nodes of different subgraphs are linked with a namic communities in real systems. We are then able to gen- probability pout. Such probabilities match the link densities within and erate three standardized benchmarks: one consists in communities which grow and shrink in size (keeping ﬁxed the total between subgraphs. We deal with three general evolutionary scenarios number of nodes of the network), while the second considers communities that merge and split. The third one is a mixed t=0 t=τ/2 t=τ that can be combined in several ways: Schematic representation of version of the previous two, and consists of a combination of (c) Mixed the benchmarks. (a) Grow/Shrink benchmark with q = 2. We begin the last four operations. with two equal- sized communities, and over a period of τ nodes

A. Grow/shrink benchmark move from the bottom community to the top, then from the top to bottom, then back to the symmetric state. (b) Merge/Split This process models the movement of nodes from one community to another. At all times, two communities are kept in benchmark with q = 2. We begin with two communities, and over a aSBMensemblewithintra-communitylinkdensitypin and inter-community link density pout.However,thenumberof t=0 t=τ/4 t=τ/2 t=3τ/4 t=τ period of τ we linearly add edges until there is one community with nodes in the two communities changes over time. In the basic uniform link density, then reverse the process. (c) Mixed benchmark with q process, we have a total of 2n nodes in two communities. In FIG. 1: (Color online) Schematic representation of the benchmarks. the balanced state, these are split into two equal communities (a)= Grow/Shrink 4, combining benchmark the with q merging=2.Webeginwithtwoequal- and growing processes. It is useful for benchmarking purposes in testing the of n nodes, that we call A and B. At the extremes, a fraction sized communities, and over a period of τ nodes move from the bot- f of nodes in community A will switch to community B. If tomability community of tocommunity the top, then from detection the top to bottom, algorithms then back to track properly the structural evolution. We have also introduced =2 we take n as the size of community A, then the number of to the symmetric state. (b) Merge/Split benchmark with q .We l begin with two communities, and over a period of τ we linearly add nodes in the community B is .Then,attime extended time- dependent measures for the comparison of different partitions in the dynamic case, which nr =2n − nl t edges until there is one community with uniform link density,then the number of nodes in community A is reverseallow the for process. the (c) observation Mixed benchmark with of q differences=4,combining between the outcome of the algorithms and the planted partitions the merging and growing processes. nA = n − nf [2x(t + τ/4) − 1] (1) through time. Our code for benchmark generation and the time-dependent comparison indices is available at with the τ/4 phase factor specifying equal sized communities (https://github.com/rkdarst/dynbench) and is released under the GNU General Public License. thus the ensemble is maintained. Conveniently, all edges can at t =0.Thefunctionx(t) is the triangular waveform be pre-computed and stored to allow a strictly repeating pro- 2t∗ 0 ≤ t∗ < 1/2 cess,1.3.3 with Community the state at time t being structure identical to the from state at timemulti-level network maps x(t)= ∗ ∗ (2) t + τ,inanalogytothemergingprocess. ! 2 − 2t 1/2 ≤ t < 1 Aspecialcasewhichweneedtocopewithisthesituation ∗ (with t ≡ (t/τ + φ)mod1)whichcontrolsthetimepe- where f is very high and pin is very low. When this happens, riodicity. The constant φ is a phase factor with φ =0 acommunityshrinkstoomuchanditmaybecomediscon- for the q =2case, and specified otherwise in the case of nected. In order to preserve the ensemble, we do not take q>2.Withthisformulation,wegetcommunitiesofsizes actions to totally eliminate this possibility, but we ensurethat (n, n), (n − nf, n + nf), (n, n),and(n + nf, n − nf) at n(1 − f)pin ≫ 2 to reduce the probability of disconnection. 1 2 3 However, if a disconnection occurs, the process is aborted and t/τ mod 1 = 0, 4 , 4 ,and 4 respectively. In practice, all 2n nodes are sorted in some arbitrary order, and the first nA nodes re-run. are put into community A, and the others into community B. Fig. 1(a) is a sketch of the grow/shrink benchmark for the Say these nodes are i =0to i =2n − 1. case q =2. After the community sizes are decided, the edges must be placed, taking into account that it is necessary that we keep the two communities in the proper SBM ensemble with equal B. Merge/split benchmark and independent link probability of pin at all times. The independence of pairs provides a hint on how to do this. When This process models the merging of two communities. In anodej is moved from community A to B, all the existing this setup, we have a set of 2n nodes, divided into two com- edges of node j are removed. Then, an edge is added between munities of n nodes each. Each of the two initial communities j and each node in the destination community B with equal has a link density of pin,wherethoselinksareplacedatini- and independent probability pin and between j and each node tialization and kept unmodified over time. The extreme states in community A with equal and independent probability pout, are two: the unmerged and the merged state. In the unmerged A Two-mode system B Multilayer network representation C Overlapping communities Code length = 2.53 Modularity = 0.41 Mode 1 Sampling Partition 1

Clustering Code length = 2.77 Modularity = 0.60 Mode 2 Partition 2 Sampling

Overlapping communities in multilayer benchmark networks. We generate the multilayer networks in two steps. (A) First we generate T LFR benchmark networks with well-defined communities, here illustrated with two network modes in blue and red. (B) Then we sample L network layers from each mode network, here illustrated with four network layers in total. (C) Each state node in the multilayer network is classified in a community, such that communities of physical nodes map overlap. In partition 1, each state node is correctly classified. In partition 2, however, the light and dark color shades are assigned to the same module, respectively

Task T1.4 Multi-level Percolation and Critical phenomena 1.4.1 Percolation in networks that are both Interconnected and Interdependent Introducing interactions between networks is analogous to introducing interactions among molecules in the ideal gas model. Interactions among molecules lead to the replacement of the ideal gas law by the van der Waals equation that predicts a liquid-gas first order phase transition line ending at a critical point characterized by a second order transition. Similarly, interactions between networks give rise to a first order percolation phase transition line that changes at the critical point to a second order transition, as the coupling strength between the networks is reduced. In this task we extend our study of interdependent networks. Specifically, we focus on the following aims: • Understanding percolation in networks that are both interconnected and interdependent • Understanding of networks of networks

Subtask 1.4.1 Percolation in networks that are both Interconnected and Interdependent Real networks continuously interact with each other, composing a large complex system, and, with the enhanced development of technology, the coupling between many networks becomes more complex and more significant. Until now two different types of coupled networks models have been suggested, (i) "Interconnected networks" where, in addition to the links connecting the elements within a network, there are also links connecting different networks, and (ii) "Interdependent networks" where nodes in one network depend on nodes in the other network for functionality. However, real coupled networks often contain both types of links, interdependent as well as interconnected links, which will be considered in this subtask. The important characteristic of such systems is that a failure of nodes in one network carries implications not only for this network but also for the function of other dependent or connected networks. In this way it is possible to have cascading failures between the coupled networks that may lead to a catastrophic collapse of the whole system. Nevertheless, small clusters disconnected from the giant component in one network can still function through interconnected links connecting them to the giant component of the other network. Thus, we have two competing effects; the interconnectivity links increase the robustness of the system, while the interdependency links decrease its robustness. Here we aim to study the competition of the two types of interlinks on the system robustness using a percolation approach. 1.4.2 Percolation on Network of Networks (NON) The approach is based on analyzing the dynamical process of the cascading failures. Our approach for NON will generalize the percolation properties of a single network (n=1). Our preliminary results indicate that while for n=1 the percolation transition is a second-order transition, for n>1 cascading failures occur and the transition becomes first order. These results for n interdependent networks suggest that the classical percolation theory extensively studied in physics and mathematics is only a limiting case of n=1 of a general theory of percolation in NON. The general theory that will be developed here is expected to have many features that are not present in the classical percolation theory. 1.4.3 Interdependent spatially embedded networks Most real networks are embedded either in two- or in three-dimensional space and the nodes in different networks are usually interdependent with each other. For instance, the power used by a computer in a computer network is dependent upon the functioning of a local power grid network. Within this task, we got the following results: • We developed a method to generate spatially embedded SF networks with a power law distribution of link lengths characterized by the spatial exponent δ [Emmerich PRE 2014]. • Analytic derivation of effects of partial dependency on interdependent spatially embedded networks with random dependency links [Bashan Nature Physics 2013]. • Determination of effects of partial dependency in spatially embedded networks with spatially constrained dependency links. [Danziger SITIS 2013, Danziger J Complex Networks 2014]. Task T1.5 Information Theory and null models of multilevel complex networks (MLCNs) 1.5.1 Information-theoretic formalism for MLCNs We have studied maximum-entropy ensembles of single-layer and multi-layer networks within an information-theoretic framework and documented a previously unnoticed breaking of the expected equivalence between ''canonical'' and ''microcanonical'' ensembles. This is a deep result questining some of the basic assumptions in statistical physics, also pointing out a different ''information content'' of the two ensembles. We have provided mathematical proofs of non-equivalence and an exact quantification in terms of local node properties. This result has practical consequences as it implies the need of a principled choice of ensembles in the analysis of real-world multiplexes. 1.5.2 Information profiles of MLCNs We extensively studied the problem of reconstruction of multi-layer and single-layer networks from partial information. In this respect, it should be noted that even a single-layer weighted network can be represented as a 2-layer multiplex where one layer contains the weighed links, and the other layer contains the binary (0/1) projection of such links. Although this representation might look trivial, it becomes quite relevant in many practical situations where one has incomplete information about both layers, and wants to recover the structure of the original network reliably. In particular, we have provided ways to estimate realistic connection probabilities (in the binary layer) and link weights (in the weighted layer) from aggregate weighted information (node strengths) and progressively more limited binary information (all node degrees, partial node degrees, total number of links, partial link density). Alternatively, we have replaced the knowledge of structural information with non-topological quantities (or ''fitness parameters''), e.g. the GDP in the case of international trade.

1.5.3 Null models of MLCNs We have devised an information-theoretic maximum-entropy method to define null models of real-world (as opposed to synthetic) multi-layer networks. The method proceeds by decorrelating the different layers of a multiplex while keeping the empirical node-specific properties (e.g. degrees, strenghts, reciprocated degrees, etc.) of all the nodes in all the layers fixed to the empirical ones. Our codes are publicly available via download from MathWorks [http://mathworks.it/matlabcentral/fileexchange/46912-max-sam-package-zip]. Since these null models assume independent layers, their definition can be carried out at the level of individual layers. We have therefore focused on the problem of sampling random graphs with given properties. Indeed, this is a key step in the analysis of many (multiplex) networks, as random ensembles represent basic null models required to identify patterns such as communities and motifs, or multiplexity itself.

1.5.4 Information loss in MLCNs The gravity model satisfactorily predicts the volume of trade between connected countries, but cannot reproduce the missing links (i.e. the topology). On the other hand, the fitness model can successfully replicate the topology of the ITN, but cannot predict the volumes. We made an important step forward in the unification of those two frameworks, by proposing a new GDP-driven model which can simultaneously reproduce the binary and the weighted properties of the ITN. Specifically, we adopt a maximum-entropy approach where both the degree and the strength of each node are preserved. We then identify strong nonlinear relationships between the GDP and the parameters of the model. This ultimately results in a weighted generalization of the fitness model of trade, where the GDP plays the role of a ‘macroeconomic fitness’ shaping the binary and the weighted structure of the ITN simultaneously. Our model mathematically explains an important asymmetry in the role of binary and weighted network properties, namely the fact that binary properties can be inferred without the knowledge of weighted ones, while the opposite is not true

1.5.5 Improving community detection for MLCNs In order to perform the community detection leading to the result discussed above, we preliminary had to redefine communities for correlation matrices, as opposed to networks. Indeed, (weighted) networks and correlation matrices have very different mathematical properties, and these differences should be taken into account when defining the null models required to identify communities. We also considered the problem of community detection in multilayer networks and approached it from the point of view of many-objective optimization, i.e. the simultaneous maximization of many pairwise modularity functions that can potentially lead to conflicts among pairs of layers.

1.5.6 Percolation on dense MLCNs We have used the method 1) to show that the striking heterogeneity of node properties in real-world multiplexes makes ordinary measures of inter-layer overlap (or correlation) inappropriate. In particular, even when layers are truly independent of each other, a large but spurious overlap and/or correlation would be measured. For this reason, we have introduced new multiplexity metrics that filter out the spurious effects of node heterogeneity and provide a more faithful quantification of inter-layer coupling. In the case of directed networks, this approach led us to the analysis of the novel property of ''multireciprocity'', i.e. the tendency of directed links in one layer to be reciprocated by directed links in a different layer.

Task T1.6 Robustness of resource distribution on networks 1.6.1 Application of Potts-like model in simple networks Already in the 70’s it has been shown that bond percolation can be formulated in terms of the Potts model by using an appropriate mapping in the Hamiltonian variable. Here we show how this method can help in solving the problem of color avoiding percolation 1.6.2 Optimization with Potts-like functional in simple networks Color-avoiding percolation provides a theoretical framework for analyzing this scenario, focusing on the maximal set of nodes that can be connected via multiple color-avoiding paths. We extend the basic theory of color-avoiding percolation that was published in [Krause et. al., Phys. Rev. X 6 (2016) 041022]. We explicitly account for the fact that the same particular link can be part of different paths avoiding different colors. This fact was previously accounted for with a heuristic approximation. 1.6.3 Application of Potts-like model in multi-level networks The critical behavior of color avoiding percolation has a rich, multi-faceted nature, depending on graph topology and multilevel structure, number of avoided colors and color frequencies. We discussed a number of phenomena using special cases. We found a transition point that can be understood within the general theory framework, by referencing standard percolation. Work package 2 Foundational study of Dynamics of Multi-level Complex Systems Task T2.1 Describing the Dynamics of coevolving interdependent Multi-level Networks 2.1.1 2.1.2 Exogenous/Endogenous Dynamics on Networks In our first attempt to define a new measure of centrality for dynamical networks, we focused on the arrival time distributions of an exogenous dynamical process such as an epidemic spreading process. When computed in terms of “activity clocks” inherent to each node of the network, these distributions have been shown to exhibit a very robust behavior. We define hierarchies of null and generative models of time-varying networks (Fig. 1) and show that the empirical patterns can be understood in terms of heterogeneous inter-event time distributions coupled with heterogeneous numbers of events per edge. We also showed, both empirically and based on a synthetic dataset, that significant deviations from the standard behavior can be caused by the presence of edge classes with strong activity correlations. Fig 1. Shuffling procedures that selectively erase specific correlations in a time-varying network, for the simple case of a network with four nodes (A, B, C, D) and two links (A– B and C–D) with their respective link activity time series. 2.1.3 Networks from Agents activities In NCGs, nodes can decide individually which edges they want to buy in order to minimize their private costs, i.e., the costs of the bought edges plus costs for communicating with other nodes. Each node v can buy a set of edges, each for a price α > 0. Its goal is to minimize its private costs, i.e., the sum (SUM-game) or maximum (MAX-game) of the distances from v to all other nodes in the network plus the costs of the bought edges. Since all decisions are taken individually and only with respect to optimize their private costs, analyzing the resulting network by comparing it to an overall good structure constitutes the central aspect in the study of NCGs. This task was formalized as analyzing the price of anarchy and was first discussed by Fabrikant et al. for the SUM-game and by Demaine et al. for the MAX-game. These papers inspired a series of subsequent works.

2.1.4 Temporal robustness of changing networks This task has been made in close collaboration with work done in Task 1.1. In particular we studied a dynamic extension of the model of interdependent networks and introduce the possibility of link formation with a probability w, called healing, to bridge non- functioning nodes and enhance network resilience. A single random node is removed, which may initiate an avalanche. After each removal step healing sets in resulting in a new topology. Then a new node fails and the process continues until the giant component disappears either in a catastrophic breakdown or in a smooth transition. Simulation results are presented for square lattices as starting networks under random attacks of constant intensity. We find that the shift in the position of the breakdown has a power-law scaling as a function of the healing probability with an exponent close to 1. Below a critical healing probability, catastrophic cascades form and the average degree of surviving nodes decrease monotonically, while above this value there are no macroscopic cascades and the average degree has first an increasing character and decreases only at the very late stage of the process. In other words at the critical healing the transition related to the vanishing giant component changes from first order to second order. These findings facilitate to plan intervention in case of crisis situation by describing the efficiency of healing efforts needed to suppress cascading failures.

Part of Network A of a simulated system at p = 0.7 at a) no healing (w = 0.0) b) below the critical healing (w = 0.2, the average degree stays below 4) and c) slightly above the critical healing (w = 0.4). d) This latter w = 0.4 system is also represented at p = 0.2 where one can observe that the nodes get more and more connected and the healing process establishes links between distant nodes.

Task T2.2 Maintaining network properties under dynamics 2.2.1 Maintaining observability and controllability In this subtask, the overall goal is to provide tools to preserve both observability and controllability of the system. Here, the dynamic is mainly exogenous and we try to observe certain aspects of the state of the network or influence the network to reach a certain goal. We consider that each node discovers all other participants in a dynamic environment with minimal communication costs. We model the knowledge of the nodes as a virtual overlay network given by a directed graph such that complete knowledge of all nodes corresponds to a complete graph in the overlay network. Although there are several solutions for resource discovery, our solution is the first that achieves worst-case optimal work for each node, i.e. the number of addresses (O(n)) or bits (O(n log n)) a node receives or sends matches the lower bound, while ensuring only a linear runtime (O(n)) on the number of rounds. We introduce a general class of strategic games that we term routing games with progressive filling. In such a game, there is a finite set of resources and a strategy of an agent corresponds to a subset of resources. Resources have capacities and the utility of every agent equals the obtained bandwidth that in turn is defined by a predefined water-filling algorithm 2.2.2 Maintaining scalability and robustness In our work on “Network Creation Games with Various Edge Qualities”, we introduce heterogenous edge qualities and explore how selfish agents that modify a network in order to optimize their utility can reach stable states and compare the cost of such a solution to the optimal solutions. In our work on “ Ca-Re-Chord: A Churn Resistant Self-stabilizing Chord Overlay Network”, we deal with a self-stabilizing process that after leaving and joining nodes reestablishes the structure of a Chord network. In our work on “Sublinear-Time Maintenance of Breadth-First Spanning Tree in Partially Dynamic Networks”, we study how a distributed algorithm can maintain connectivity information in form of a breadth-first spanning tree when the network changes either only by edge insertions or by only edge deletions. In our work on “Reactive Planar Spanner Construction in Wireless Ad Hoc and Sensor Networks”, we provide an important tool for stateless routing which is suitable for highly dynamic environments. 2.2.3 Certifying distributed algorithms Certifying distributed algorithms are a helpful tool to determine whether some property of the network is satisfied. Especially, local certificates are useful in dynamic systems since they can be checked quickly. In this subtask, we want to explore the possibilities and limitations of approaches that use of certificates in order to observe the network and support the first two subtasks. In our work on “ Hierarchies in Local Distributed Decision”, we give a complexity theoretic characterization of what certifying algorithms can achieve if the size of the certificates is restricted. In classical theory of computation, certain problems are supposed to be harder to be solved by Turing machines compared to other problems. For example, given a Boolean function, the Boolean satisfiability problem is to determine if its input variables can be assigned in such a way to make the formula evaluate to true. While it is supposed to be hard to solve this problem (namely NP-hard) , it is rather easy to verify if a formula is satisfiable given a valid input assignment. A valid input assignment serves as a certificate for the problem. This idea of certification can be transferred to distributed computing.

Task T2.3 Risk Assessment and Management in Complex Networks 2.3.1 Developing Notions of Risk Here, we report the initial study on the rare event in networks. One of the model examples of rare events in networks is a rumor spreading process. It is well known that the event that some rumor generates a cascading effect and spreads to many nodes in a network is very improbable. Our study focused on two main issues. On one hand we were working on devising exact and efficient models for the description of rumor spreading. Such models will be the starting point for the further more detailed analysis of the processes. Observer, that he efficiency of models is the key aspects for such studies, because in order to see rare events one needs to perform huge number of simulations. 2.3.2 Classifying Rare Events in Networks On the other hand, our experimental efforts so far have concentrated on gathering rumor spreading data from different sources. However, the main problem that we have been faced with is the vast amount of such data, which makes finding actual rumors very hard. We are now working on new algorithmic methods that would allow for automatic recognition of rumour spreading events. 2.3.3 Prediction and Reaction to Rare Events Although prediction for a stochastic graph models is interesting, our main goal is to show that accurate forecasting is possible for real life social networks. As we were able to demonstrate, the degree distribution of a single, specific social network, is all that is needed to predict with very good accuracy the speed at which rumour spreading evolves. In other words, we observe that average behavior is typical for social networks. Quite surprisingly, in our opinion, we are not only able to predict how fast the rumor reaches every node, but also estimate well the average number of informed nodes in each round

Task T2.4 Controllability and Network Dynamics 2.4.1 Quantifying the “power” of an individual node in controlling the full system We formalize the dynamics of a controlled network with N nodes and ND external control inputs as

x!(t) = Ax(t) + Bu(t) , where the vector x(t) = [x1(t), x2(t), . . . , xN(t)] describes the states of the N nodes at time t. In (1) xi(t) can represent the concentration of a metabolite in a metabolic network, the geometric state of a chromosome in a chromosomal interaction network, or the belief of an individual in opinion dynamics. The vector u(t) = [u1(t),u2(t),...,uND(t)] represents the external control inputs, and B is the input matrix with Bij = 1 if the control input uj(t) is imposed on node i. The adjacency matrix A captures the interactions between the nodes, including the possibility of self-loops Aii representing the self-regulation of node i. If at t = 0 the system is in state xo = 0, the minimum energy required to move the system to point xd in T −1 the state space can be shown to be E(τ ) = xd G (τ )xd ,

τ T where G(τ ) = e At BB T e A t dt is the symmetric controllability Gramian. When the system is controllable ∫0 all eigenvalues of G(τ ) are positive. Equation indicates that for a network A and an input matrix B the control energy E(τ ) also depends on the desired state xd. Consequently, driving a network to various directions in the state space requires different amounts of energy. The minimum energies are shown below The control energy surface for all normalized desired states is an ellipsoid, implying that the required energy varies dramatically as we move the system in different directions.

2.4.2 Matching the mathematical tools with real data/2.4.3 Spectral properties of multi-level networks

The key result we have obtained is that the variability of control energy and observation uncertainty for different directions of the state space depends strongly on the number of driver nodes. Specifically, if all nodes are directly driven, the distribution p(E) of eigen-energies is uniquely determined by the network topology. For example, for a scale- free network the distribution p(E) is also a power law with the same exponent as that of the degree distribution; for non-scale-free networks the distribution p(E) is bounded. Thus, if all nodes are directly driven, there are no significant energetic barriers for control. If we drive only one node, the p(E) distribution is a power law with exponent -1. This implies that the required energy varies enormously for different directions, independent of the network topology. In this case the energy corresponding to the most difficult direction increases exponentially with the system size, thus control in certain directions is energetically prohibitive for large networks. This result sheds light on to the recent debate about whether a single driver node can fully control the whole network. Our results offer a simple answer: Yes in principle, but not in practice. If a finite fraction of nodes are directly driven, the largest energy will decrease exponentially when the number of driver nodes increases. This result has significant implications for the practical trade-off between the number of controllers and the cost of control.

Task T2.5. Mechanism Design in Complex Networks 2.5.1 Controllability by Mechanism Design Resilience, a system’s ability to adjust its activity to retain its basic functionality when errors, failures and environmental changes occur, is a defining property of many complex systems. Despite widespread consequences for human health, the economy and the environment, events leading to loss of resilience — from cascading failures in technological systems to mass extinctions in ecological networks — are rarely predictable and are often irreversible. We bridge this theoretical gap by developing a set of analytical tools with which to identify the natural control and state parameters of a multi-dimensional complex system, helping us derive effective one-dimensional dynamics that accurately predict the system’s resilience The proposed analytical framework allows us systematically to separate the roles of the system’s dynamics and topology, collapsing the behaviour of different networks onto a single universal resilience function.

2.5.2 Observability by Mechanism Design Static networks are known to display energetically prohibitive final states, i.e. states that are accessible to the system, but require exorbitant amount of energy to reach them. We found that the average control energy of temporal network is many orders of magnitude smaller than that of the corresponding static network, a difference particularly large for small time window. This indicates that high temporality, in the form of rapid switches in the network topology, allows for considerable savings in the control energy. In both temporal and static networks we can reduce the control energy by using more driver nodes, but the gap between the temporal and static network persists until all nodes are directly controlled. In this case we do not need to exploit the time-varying topology to cheaply move between states, as the state of each node can be dictated directly.

Task T2.6 Dynamics of recovery and flow in multi-level networks 2.6.1 Topological Dynamics

BOOTSTRAP PERCOLATION IN MULTIPLEX NETWORKS We introduced an activation model on multiplex networks, inspired by the bootstrap percolation process on a simplex network. The goal is to represent activation of vertices on a network, such as in social mobilisation, or the repair of infrastructure networks after a disaster. We also define its counterpart pruning process. The models are bootstrap percolation (WBP) and weak pruning percolation (WPP). These are natural extensions of percolation on simplex networks, and are somewhat analogous to bootstrap percolation and the k-core pruning algorithm on simplex networks. We have shown that, unlike the case of a single layer, these two models are distinct and give origin to different critical behaviors. We further introduced the concept of invulnerable nodes in multiplex percolation, and shown their effect on the critical transitions associated with the emergence of a giant percolating cluster. We have explicitly calculated the critical phenomena characterizing multiplex networks made of Erdös-Rényi networks on each layer. The WBP model includes both continuous and discontinuous hybrid transitions for two or more layers, while the WPP model has only continuous transitions in 2 layers, but a discontinuous hybrid transition appears when there are 3 or more layers. The discontinuous transition in the 3-layer WPP disappears at the same critical point as that in 2-layer WBP. The WPP model, besides filling a gap in our understanding of percolation on multilayer networks, also provides an interesting diagnostics to evaluate the presence of a missing layer. The absence of critical points and discontinuous transitions in the 2-layer case, in fact, is remarkable as it qualitatively differs from the 3-layer case, and it might be used to determine the layer structure of multiplex networks with limited information. The WBP model constitutes the simplest activation process that may occur on a multiplex. Differently from other more complex models, we provided a simple analytical method, which allows not only exact calculation of the critical behavior in locally tree-like networks, but also the calculation of critical exponents and the characterization of critical clusters.

INTERDEPENDENT NETWORKS WITH INTERSIMILARITY Real interdependent networks are sometimes coupled according to some intersimilarity features. Intersimilarity means the tendency of neighboring nodes in one network to be interdependent of neighboring nodes in the other network. Such coupled networks are more robust against cascading failures than randomly coupled interdependent networks. We introduced a method to analytically calculate the cascading process of failures in interdependent networks with common links. To analyze this problem, we considered the cluster components of the network composed of only common links after the initial attack. All nodes in such a component will survive or fail simultaneously during the cascading process. Based on this fundamental feature, we divided the system into subnetworks according to the sizes of the components, and then contract all nodes in each component into a single node. After contraction, the system degenerates into two randomly coupled networks without common links, which is solved analytically. We found the exact solution for the case where A and B are fully interdependent Erdös-Rényi (ER) networks each of average degree k and the network of common links is also an ER network with average degree k. In this case we showed that the interdependent networks system undergoes a first order transition for all k ≥ 0.

2.6.2 Dynamics of information flow in networks of networks We studied percolation (i) in two partially interdependent networks with clustering within each network and (ii) in a network of clustered networks (NON), i.e., a network consisting of more than two interdependent clustered networks. We studied how clustering within the networks influences such percolation properties as the critical threshold pc at which the giant component collapses, the sizes of the giant components ψ∞ and φ∞ in the two networks, the critical coupling qc at which the first-order phase transition changes to a second-order phase transition, and the dynamics of cascading failure between two clustered networks. For each clustering coefficient, the system shows a first-order to second-order transition as we decrease coupling strength q. As we increase the clustering coefficient of each network, the system becomes less robust. Such influence of the clustering coefficient on network robustness decreases as we decrease the coupling strength, and the critical coupling strength qc, at which the first-order phase transition changes to second-order, increases as we increase the clustering coefficient. We have also investigated the differences and similarities between two different joint degree distributions, the fixed-degree distribution (FDD) and the double Poisson distribution (DPD). We have found that, although the percolation threshold is different in the two cases, the general conclusion that an increase in the clustering coefficient causes interdependent networks to become less robust holds. Simulation results agree well with theoretical results in all cases.

Work package 3 Modelling of Multi-level Complex Systems Task T3.1: Agent-based modelling and game-theoretic evolutionary dynamics in complex networks 3.1.1 Evolutionary dynamics in complex networks When studying evolutionary dynamics in complex networks, the underlying static or dynamic graph imposes serious restrictions on interactions: Only neighbor nodes (agents) can interact at a given time. In addition, the structure of the graph may be hierarchical or periodic (multi-level) and/or the frequency of mutation and even of selection may vary (different time scales). Our research in this area includes the investigation of new directions of research on evolutionary dynamics in complex networks emerging from well-known evolutionary models, such as new mutation mechanisms and multiplicity of types of individuals on the Moran process, multi-scalability of networks, rules of interaction among neighbors, new notions of stability in multi- scale dynamics on networks with local interactions, the effect of heterogeneity of population members on their survivability, the effect of the network structure and the scale of interactions among neighbors on the long term results of the dynamics. 3.1.2 Agent-based game-theoretic co-evolution mechanisms The modeling effort in this subtask concerns capturing systemic properties of co-evolving systems on time- varying networks. Interaction topology is captured by a type of “network Hamiltonian”, which allows 3 treating all possible interaction topologies in a harmonized unified way. The aim of this subtask is to probe to what extent it is possible to use the concept of phase transitions and consequently phase diagrams in coevolving systems, where the full Hamiltonian governs the interactions (between strategies of agents) and at the same time the interaction topology, which results as an emergent property. Furthermore, in this subtask we explore under which conditions certain typical network features (scale free degree distributions, non-trivial clustering, etc.) emerge as a result of the extended Hamiltonian. 3.1.3 Modeling and simulation platform for public use. Except for very simple states of agents, like for example Ising spins, these systems are extremely hard to solve analytically. Therefore, we also rely on computer simulations. We provide the code of some of our models. In particular we can summarize the contributions regarding Social Contagion. Specifically: 1. Modeling self-sustained activity cascades in socio-technical networks. Here we present a mechanistic model that accounts for the temporal evolution of the individual state in a simplified setup. We model the activity of the individuals as a complex network of interacting integrate-and- fire oscillators. The model reproduces the statistical characteristics of the cascades in real systems, and provides a framework to study time-evolution of cascades in a state-dependent activity scenario. 2. Contact-based Social Contagion in Multiplex Networks(E. Cozzo, R. A. Baños, S. Meloni, and Y. Moreno, Contact-based social contagion in multiplex networks, Physical Review E 88, 050801(R) (2013). We develop a theoretical framework for the study of epidemic-like social contagion in large scale social systems. The framework is applied to a number of different situations, including a real multiplex system. Finally, we also show that when the system through which information is disseminating is inherently multiplex, working with the graph that results from the aggregation of the different layers is inaccurate. 3. Quantyfing long-term scientific impact(Wang, D., Song, C., and Barabási, A.-L. Quantifying long- term scientific impact. Science (2013). We develop a of predictability of citation-based measures frequently used to gauge impact, from impact factors to short-term citations. Here, we derive a mechanistic model for the citation dynamics of individual papers, allowing us to collapse the citation histories of papers from different journals and disciplines into a single curve, indicating that all papers tend to follow the same universal temporal pattern. The observed patterns not only help us uncover basic mechanisms that govern scientific impact but also offer reliable measures of influence that may have potential policy implications.

Task T3.2 Modeling economic Multi-level and Multi-Scale Networks 3.2.1 Modeling economic links In order to refine our understanding of multi- level systems we addressed the analysis of financial distress propagation in multi-level networks from real data. This work is the result of cooperation between the Medical University of Vienna and the Banco de Mèxico. We are using a complete set of financial transaction data to quantify the marginal effect on expected loss of individual interbank transactions for the whole economy and for individual banks.

We also used network models to describe the evolution of a particular economic system, namely the International Trade Network (ITN). Previous studies often assume that globalization and regionalization in international trade are contradictory to each other. We re-examine the relationship between globalization and regionalization by viewing the international trade system as an interdependent complex network. We use the modularity optimization method to detect communities and community cores in the ITN during the years 1995–2011. We find rich dynamics over time both inter- and intra-communities. In particular, the Asia- Oceania community disappeared and reemerged over time along with a switch in leadership from Japan to China. We provide a multilevel description of the evolution of the network where the global dynamics (i.e., communities disappear or reemerge) and the regional dynamics (i.e., community core changes between community members) are related. Moreover, simulation results show that the global dynamics can be generated by a simple dynamic-edge-weight mechanism.

3.2.2 Modeling economic aggregation

The R&D network is constructed by linking two firms every time an alliance is announced in thedataset. When an alliance involved more than two firms (consortium), all the firms involved were connected in pairs, resulting into a fully connected clique. Multiple links between the same firms are allowed (two firms can have more than one collaboration on different projects), while isolated firms i.e. firms that had no alliances in a given period, are not part of the network

the pooled R&D network snapshots in 1989, 1993, 1997, 2001 and 2005. We plotted – in different colours – only the ten largest sectors, in order to ease visualisation. The picture denotes the presence of different phases in the evolution of the R&D network. More precisely, the plots suggest the presence of a significant network growth until 1997, and a reversal of this trend in the last periods of our sample.

Task T3.3 Modeling Information Diffusion and Navigability of Complex Networks 3.3.1 The algorithmic and mathematical foundations of rumor spreading The network inference problem consists of reconstructing the edge set of a network given traces representing the chronology of infection times as epidemics spread through the network. This problem is a paradigmatic representative of prediction tasks in machine learning that require deducing a latent structure from observed patterns of activity in a network, which often require an unrealistically large number of resources (e.g., amount of available data, or computational time). A fundamental question is to understand which properties we can predict with a reasonable degree of accuracy with the available resources, and which we cannot. We define the trace complexity as the number of distinct traces required to achieve high fidelity in reconstructing the topology of the unobserved network or, more generally, some of its properties. We give algorithms that are competitive with, while being simpler and more efficient than, existing network inference approaches. Moreover, we prove that our algorithms are nearly optimal, by proving an information-theoretic lower bound on the number of traces that an optimal inference algorithm requires for performing this task in the general case.

MODEL OF OPINION DIFFUSION: TWITTER ANALYSIS OF ELECTIONS In this case we made an analysis of the behavior of Italian Twitter users during national political elections. We monitored the volumes of the tweets related to the leaders of the various political parties and we compareed them to the elections results. The analysis was performed both at a national and regional levels, ranging from the regions composing the country to macro-areas (North, Centre, South). We followed the state of the art approach in the analysis of the Twitter data. That is, we considered both the volume of the tweets and the ratio of the time variation of this volume when considering two different parties. The presence of candidates/parties in the press is believed to be a good indicator of the rank they will achieve in the elections. The extension of this feature to Facebook and Twitter has been attempted, but the number of studies on this topic is not large enough to obtain a clear indication. Regarding the time series of tweets, some studies seem to confirm that Twitter may be considered a good proxy for election results. In terms of the ratio of volume variation, we used a newly introduced approach known as the Relative Support (RS) parameter. The RS is an instant indicator of the comparative strength of two political parties - A and B - in Twitter, and it is given by

RS= mA/mB where mA and mB are the slopes for the accumulated mentions of the A and B political parties. For the purposes of this analysis we break Italy down into three macro-areas (North, Centre and South); the size of these geographic divisions ensures a good statistical outcome. It is worth noting that the ranking of the main parties differs in these three macro-areas with respect to the national one. Nevertheless, the vote-tweet correlation still holds at the macro-area level. The notably good results of the Lega party (Small) in the North, and the PD in the Centre are confirmed by tweet volume. This work confirms what has been already found in a series of similar analyses. Twitter data can be an effective way to get indications on the election outcomes.

3.3.2 Cascading effects and Influence Spreading we study the pricing strategies with limited time and limited price discrimination, specifically k-PD strategies and k-PR strategies. (a) In k-PD strategies the product is offered to the clients in k rounds such that (i) each client belongs to exactly one round, (ii) only the clients that have purchased the product in a previous round can influence the valuation functions of the clients to whom the product is offered in the current round, and (iii) each client has a personalized price. (b) In k-PR strategies the product is offered to the clients in k rounds such that (i) and (ii) from above hold, and (iii) all clients in the same round are offered the same price. This setting models the scenario where all clients to whom the product is offered at the same time receive the same price, for instance, when giving the product at a discount price to early adopters. Our results are: A There is a log n-PR strategy that achieves a constant factor approximation of the optimal revenue bR. Thus only log n different prices are necessary. B We show that the result of (A) is not achievable with o(log n) rounds, even for valuations with very simple externalities. C We show that even for k = 2 there exists a k-PD strategy whose revenue is within a constant factor of bR D Maximizing revenue in both of the above strategies is APX-hard, even for valuations with very simple externalities.

3.3.3 Navigability in social networks We extended the theory for well-known random walks to multiplex networks, and we introduced a new type of walk that can exist only in multi-layer networks. We calculated the salient descriptors of walks, i.e., the vertex occupation time and the time required to explore a given fraction of the multiplex. These results are extremely important to ascertain the behavior of dynamical processes on top of these structures. In particular: - For the first time, we formalize transition rules between vertices accounting for the multi-layer structure of the network; - We provide analytical expressions to calculate the coverage of the multiplex, i.e., the fraction of vertices visited by a random walker within a given amount of time, and we verify their goodness by comparing with massive Monte Carlo simulations; - We show how the efficiency in exploring the multiplex networks critically depends on the underlying topology of layers, the weight of their inter-connections and the strategy adopted to walk.

Task T3.4 Modeling Social Contagion dynamics in techno-social systems 3.4.1 Models of Social Contagion We developed a theoretical framework for the study of epidemic-like social contagion in large scale social systems. We considered the most general setting in which different communication platforms or categories form a multiplex network. Using a contact-based Markov chain approach we show that the critical point of the multiplex system associated to the active phase is determined by the layer whose contact probability matrix has the largest eigenvalue. The framework can be applied to a number of different situations, including a real multiplex system. We also show that when the system through which information is disseminating is inherently multiplex, working with the graph that results from the aggregation of the different layers is inaccurate. With this contribution we directly addressed the core questions behind task 3.4 of the project proving a new class of models to study spreading dynamics in social contexts. Moreover, the possibility for our framework to include multiple communication channels and a multilayer representation allows to study directly systems where different time and spatial scales are present at the same time. In addition, our work showed that a multilayer representation of this class of systems is fundamental as considering aggregated or single layer systems can lead to important errors. The SIS dynamics on this class of systems can be represented by two different infection probabilities one for intra-layer contagion βa and one for inter-layer diffusion γa while the time scale of the dynamics is dictated by the recovery rate µ. Extending the contact based Markov chain approach proposed in to multiplex systems we wrote down the discrete-time evolution of nodes' infection probability p(t) as: ! ! ! ! ! ! ! ! ! ! ! ! p(t +1) = (1 − p(t))∗(1 − q(t))+ (1 − µ)p(t)+ µ(1 − q(t))p(t) where * stands for elements' wise multiplication of two vectors, and qi(t) is the probability that node i will not be infected by any neighbor. Schematic of a 2-layer multiplex system where the contagion dynamics takes place. There are actors that take part in more than one layer (green nodes), whereas others are present only in one layer (red nodes). The dotted edges link two nodes that represent the same actor.

b1;2 is the contagion rate within the same layer

whereas g1;2 represents the probability that the contagion occurs between layers. The right panel shows a small network and its associated C and A.

3.4.2 The Spreading of Epidemics We analyze the structural properties of the networks that emerged to exchange messages about the protests, networks that helped diffuse relevant information before and after the mobilization days. We compare the size of information cascades and how the composition of the network changed during the year that separates the two mass mobilizations, paying special attention to how actors migrate across different regions of the network. We show that local changes in network structure underlie aggregated differences in how information diffused: an increase in network hierarchy goes hand in hand with a reduction in the average size of cascades. Although we cannot disentangle the effects that exogenous factors (like political fatigue) have on the online patterns we observe, we believe that our findings qualify models of information diffusion and add important nuances to the discussion of how the structure of networks relates to their function.

3.4.3 Competitive Viral marketing We showed how social welfare and revenue could be maximized when products have positive network externalities. We studied revenue-maximizing marketing strategies for copies of a digital good and bidders with unit-demand. Our results generalize the work in [45]. We consider offers to customers in a limited number of rounds. The offered product prices depend either on the customer or on the round, and only customers that accept the offer in some round influence the valuation of their friends in future rounds. We show that with O(log n) rounds and prices per round, where n is the number of customers, a constant factor of the revenue with individual prices for each customer can be achieved, and that this is not possible with o(log n) rounds. Moreover, we show that with a constant number of rounds and prices per round it is APX-hard to maximize the revenue and we give constant factor

Task TS3.5 Data, Text and Network Mining of Complex Multi-Level Networks 3.5.1 Network extraction from text We address the question of reliable and efficient construction of entity co-occurrence networks from textual sources on the web. This process is two-fold as first the entities of interest are identified, and second the significant co-occurrences are extracted. Entity identification in textual documents requires three components: (1) an ontology of the entities of interest and relevant terms, (2) gazetteers of the possible appearances of entities in the text, and (3) a semantic annotation procedure that finds and labels the entities. We calculate the links between the entities by our newly proposed significance algorithm (Popovic et al. 2014), based on a simple algebraic method and counting statistics, that can be efficiently used to extract significant co-occurrences from a real-time data stream. A benchmark model was used to generate synthetic data, on which the significance algorithm was tested and the required parameters were determined.

We investigated the relation between the networks extracted from online texts and the networks drawn from economic data. We demonstrate an application of our method by extracting a network of co-occurring countries from financial news. Furthermore, we construct a network of the sentiment associated with country co-occurrences. We compare these networks constructed from online news to other real-world countries networks representing relations extracted from geographical, financial and trade data (Sluban et al, 2015, submitted). The results show that the relation of geographical proximity overlaps to the greatest extent with the co-occurrence of countries in the news, whereas important trade relations between countries are mostly accompanied by positive sentiment between these countries in the news.

3.5.2 Inference on multi-level dynamic networks We developed new method for estimating magnitudes of external and peer influence in social networks using information on friendship network and the times of activation. The validation of the methodology is described in D4.10. The preliminary work on this topic was presented on ECCS 2014 conference as an ignite talk "Modeling voting activity dynamics in social network during December 1, 2013 referendum in Croatia". On a similar point we proposed a memory biased random walk model on a multilayer sequence

network (Antulov-Fantulin et al., 2014a), as a generator of synthetic sequential data for recommender systems. Here, the multilayer sequence network is constructed from clickstreams, sequence of items like web pages, movies, books, etc., that the user interacted with. Personalized recommender systems rely on each user’s personal usage data in the system, in order to assist in decision-making. However, privacy policies protecting users’ rights prevent these highly personal data from being publicly available to a wider researcher audience. We demonstrate the applicability of the generated synthetic data in training recommender system models in cases when privacy policies restrict clickstream publishing. The figure shows a transition probability calculation on a simple example.

Statistical estimation of source of contagion on networks We introduce a statistical inference framework for maximum likelihood estimation of the contagion source from a partially observed contagion spreading process on networks (including multilayer networks). As a main case study for epidemic source detection we use empirical bipartite temporal network of sexual contacts, where we simulate sexually transmitted disease spreading with the SIR model. With our novel Monte Carlo statistical inference methods for source detection, we are able to efficiently detect the patient- zero. In a realistic scenario of uncertainty in epidemic duration time and of uncertainty in temporal interactions orderings, we demonstrate that with an appropriate use of Bayesian inference we are still able to efficiently detect epidemic sources in temporal networks.

3.5.3 Sentiment analysis and modelling dynamics of sentiment propagation To correctly characterize the sentiment of short texts, like tweets, good classification models are required. We achieve this by constructing custom sentiment models using the SVM machine learning algorithm trained on language specific manually annotated tweets. For this purpose we collected over 2 million tweets, written in 14 languages (18 originating countries), and manually annotated their sentiment. This large resource of high quality data enables us to train and construct language specific sentiment models, which we use for reliable sentiment classification in various use cases.

Number of negative, neutral and positive annotated tweets for 18 countries.

Retweet networks, their communities and sentiment In a case study, we monitor Twitter activity on a wide range of environmental issues. From our collection of 30 million tweets we construct a retweet network. In addition to the analysis of the structural properties of the communities and the differences between their influence, we propose an approach to monitor and compare the sentiment of entire network communities towards selected topics. This approach for understanding the preferences of various groups of users, active in the environmental domain, results in perceivable differences in sentiment leaning.

Retweet network of the debate on environmental topics. Presented are nine largest communities, categorized according to their most influential users, and distinguished by colour.

Comparison of Twitter sentiment and stock price returns

We focus on the question whether the events present in the Twitter data stream are informative for the price movement of the US stock companies they mention. Initially, we have concluded that the events concerning stock market companies which receive the largest Twitter attention correspond to the quarterly earnings announcement news. Our research goal was to filter out these news events, and use sentiment analysis to investigate whether the remaining events are also informative for the direction of movement of the stock price. To answer this question we use an event study, as defined in financial econometrics. It analyzes the abnormal price returns observed during the events. The results of the analysis suggest that in the days after the peaks of Twitter activity the price return has abnormal variation in the direction of the aggregated sentiment of the Twitter users. This is evident both for earning announcements and for the remaining Twitter events.

Work package 4 Data driven validation in specific application domains Task T4.1 Validation of the models for real-world interdependent networks 4.1.1 Geographical Networks We worked on data including information of multinational firms’ networks of subsidiaries (the 3000 largest groups of the world and their 1 million direct and indirect subsidiaries in 2006, 2010). We included 2013 (about 1.1 million subsidiaries’ linkages) and we made the corresponding three databases. This multi- level urban approach will be also available for other database whose are right now in aggregation by Urban areas.

4.1.2 Networks not embedded in physical space The R&D network used for the empirical analysis described in manuscript arXiv:1304.3623, was complemented by a second network of firms that filled patents together in the US. More precisely, the firm alliance information in R&D projects (network 1) was extracted from the Thomson Reuters SDC Platinum alliances database, and the patent information (network 2) was obtained from the National Bureau of Economic Research, NBER patent database. The SDC database provides a total number of 21,572 alliance reports obtained by all the publicly announced R&D partnerships from 1984 to 2009, for all kinds of economic actors (including firms, investors, banks and universities). We associated every firm with its 4-digit Standard Industrial Classification (SIC) code, which enabled us to classify them to the right industrial sector of activity. This allows us to further divide the database into additional layers of economic activity, if needed.

Task T4.2 Validation of the models for multilevel economic networks 4.2.1 Validation of economic networks The motivation of the work on Patent networks is to determine ‘central’ sectors of the economy, i.e. those industries that are likely to cause extensive spillovers due to their importance to a broad range of other industries. A wide range of methods exists for measuring such a centrality in complex networks, such as PageRank, betweenness centrality, and a host of others. We carry out a more specific measure of centrality on the class level of the patent citation network. In Figure we show the behavior of the top five globally central classes for each of the six countries considered in 1995 and 2005 as the parameter ε in the model is varied. In all cases, we see an overall decrease in the centrality of high-ranked sectors due to the increase “teleportation” probability at every step of the random walk as e increases (which necessarily increases the measured value of the centrality for unimportant classes). The relative decline when comparing two central classes is entirely due to the topology of the citation network and is a signal of the difference between global and domestic centrality. We also find that high ranked classes in the US are robust to variations in e over a fairly wide range (up to ε ~ 0:1 compared to the more rapid drops observed for other countries), and the ordering of the top- five classes is unchanged over nearly the entire range of values of e. This is due to the overall centrality of the US economy, since the most patents are issued to the US a larger share of citations are domestic and are therefore robust to variations of ε

4.2.2 Validation of economic aggregation

The aggregation could be distinguished by activity sector. The network we obtain is the following, composed by aggregated networks of firms between World Cities by activity sector (A, B, C,...) (with a NACE nomenclature: 80 categories or 20 aggregated metacategories).

The fragmentation of production across countries has become an important feature of the globalization in recent decades can be represented either by geographical aggregation as above, or by means of global value chains (GVCs). When empirically investigating the GVCs, previous studies are mainly interested in knowing how global the GVCs are rather than how the GVCs look like. From a complex networks perspective, we use the World Input-Output Database (WIOD) to study the global production system. We find that the industry- level GVCs are indeed not chain-like but are better characterized by the tree topology. Hence, we compute the global value trees (GVTs) for all the industries available in the WIOD. Moreover, we compute an industry importance measure based on the GVTs and compare it with other network centrality measures. Finally, we discuss some future applications of the GVTs. We use the World Input-Output Database (WIOD) to compute the GVNs and the GVTs. At the time of writing, the WIOD input-output tables cover 35 industries for each of the 40 economies (27 EU countries and 13 major economies in other regions) plus the rest of the world (RoW) and the years from 1995 to 2011. For each year, there is a harmonized global level input-output table recording the input-output relationships between any pair of industries in any pair of economies. The numbers in the WIOD are in current basic (producers') prices and are expressed in millions of US dollars (The basic prices are also called the producers' prices, which represent the amount receivable by the producers.

The domestic and foreign GVTs where China's electrical equipment industry has the highest importance score in 2011. The upper and lower trees are the domestic and foreign trees respectively where China's electrical equipment industry has the highest importance score. The edge weight threshold is set to 0.01 and the number of rounds is limited to 3. Different colors of the nodes indicate different countries. The red edges indicate cross- country relationships while the blue edges indicate domestic relationships. The edge width is proportional to the edge weight, i.e., the share of the value-added contribution.

Task T4.3 Making contact between network models and measurements 4.3.1 Defining description levels for multi-scale model validation We have developed a novel tool to characterize temporal networks. Temporal aggregation is often performed in order to obtain a static summary of a temporal network. However, different temporal networks can give rise to the same aggregated network. In particular, paths that are present in the aggregated version might be authorized by causality constraints in one temporal network but not in another one. The betweenness preference introduced allows to quantify to what extent paths existing in time-aggregated representations of temporal networks, are actually realizable based on the sequence of interactions. 4.3.2 Synopsis of empirical networks in relation to dynamical processes We have defined a series of models of interacting agents, with tunable interaction rules. In particular, various interaction ingredients are based on memory effects empirically observed in human contact data, similar to rich-get-richer effects. While several memory effects are present at the same time in the data, the modelling framework allows us to analyse separately their individual roles in determining the complex phenomenology arising in the data, concerning in particular the heterogeneous distributions of contact durations and numbers and of inter-contact durations. We have hence disentangled the role of these mechanisms, shown which empirical characteristics emerge from each, and shown that the inclusion of all memory effects is necessary to recover an overall phenomenology fully resembling the one of the empirical data.

Task T4.4 Validation of the structural and dynamical framework for multi-level techno-social networks 4.4.1 Reconciling proxies and scales We collected and analyzed messages exchanged in Twitter using two of the platforms publicly available APIs (the search and stream specifications). We assessed the differences between the two samples, and compared the networks of communication reconstructed from them. The empirical context is given by political protests taking place in May 2012: we tracked online communication around these protests for the period of one month, and reconstructed the network of mentions and re-tweets according to the two samples. We found that the search API over-represents the more central users and does not offer an accurate picture of peripheral activity; we also found that the bias is greater for the network of mentions.

4.4.2 Validation of spreading processes at multiple scales On the basis of these messages, we reconstructed protest-relevant communication networks using two Twitter conventions: @mentions and RTs. The first, @mentions, target other users by explicitly mentioning their username; RTs, on the other hand, are employed to broadcast messages previously published by other users. The networks reconstructed using these two features offer a more accurate approximation to actual communication exchange than the underlying following/follower structure: this network offers the potential for information flow but that potential is realized differently in specific information domains. To reconstruct the @mentions and RTs networks, we parsed the text of the messages captured in the two samples and created a directed link between users i and j every time i mentioned or re-tweeted j. What these trends suggest is that the picture that the two samples give of the protesters (or, to be more specific, of users engaging in the exchange and diffusion of protest-related information) is, overall, quite different the coefficients never go much higher than 0.5; but it also means that the resolution of this picture changes over time, becoming less accurate when there is a lower volume of information exchange, as it happened after the demonstrations. This suggests that the search API, which has higher limitations in data collection, might be a less appropriate tool for periods of low activity or information streams that are not very abundant. Task T4.5 Validation of data, text and network mining of complex multi-level networks The dataset consists of a lexicon of 751 emoji characters with automatically assigned sentiment. The sentiment was computed from 70,000 tweets, labeled by 83 human annotators in 13 European languages. The process and analysis of emoji sentiment ranking is described in the paper: Kralj Novak P, Smailović J, Sluban B, Mozetič I (2015) Sentiment of Emojis. PLoS ONE 10(12): e0144296. doi:10.1371/journal.pone.0144296. The online version and demo of the lexicon Emoji Sentiment Ranking 1.0 is available at http://kt.ijs.si/data/Emoji_sentiment_ranking/ . All the data are available at the Slovenian language resource repository CLARIN.SI, http://hdl.handle.net/11356/1048. 4.5.1 Validation of multi-level sentiment analysis and modelling of sentiment propagation Our approach to sentiment analysis is based on supervised machine learning. The procedure consists of the following steps: (i) a large sample of tweets is first manually annotated with sentiment by human annotators, (ii) the labeled set is used to train and tune a classifier, (iii) the classifier is evaluated by crossvalidation and compared to the inter-annotator agreement, and (iv) the classifier is applied to the whole set of collected tweets or comments on Facebook. We provide the first emoji sentiment lexicon, called the Emoji Sentiment Ranking, and draw a sentiment map of the 751 most frequently used emojis. We observe no significant differences in the emoji rankings between the 13 languages and the Emoji Sentiment Ranking. Consequently, we propose our Emoji Sentiment Ranking as a European language-independent resource for automated sentiment analysis. Finally, the paper provides a formalization of sentiment and a novel visualization in the form of a sentiment bar. Sentiment map of the 751 emojis. Left: negative (red), right: positive (green), top: neutral (yellow). Bubble size is proportional to log10 of the emoji occurrences in the Emoji Sentiment Ranking. 4.5.2 Validation of detection and description of events in multi-layered networks In order to deal with inference from data (either nodewise or edgewise) in a multiview/ multilayer setting, we significantly extended algorithms for redescription mining and generalized the process of generation and optimization of redescription sets. These are described in the context of D3.11. We have built a novel interactive exploration and visualization tool called InterSet, that accompanies our redescription mining algorithmic framework described in 3.11. The tool is developed to allow users efficient interaction with possibly large redescription sets; it enables better understanding of the structure of the redescriptions and relations between attributes of the underlying data. In InterSet insights from redescription sets can be obtained through three different interaction modes based on: (i) similarity of entity occurrence in redescription support sets, (ii) attribute co-occurence in redescriptions and (iii) redescription quality measures. These modes provide necessary contextualization, which is a major advantage compared to current state of the art approaches. 4.5.3 Augmenting multi- level networks with networks extracted from tex InterSet provides means to filter out subsets of redescriptions based on user interaction in three views, which user can analyse further on the level of individual redescriptions. In our work we demonstrate the use of the InterSet on country data, where one view of attributes are attributes describing different demographic and economic country parameters while the other view encompasses import/export relations of countries.

Task T4.6 Socio-semantic multi-level networks and the emergent complexity of scientific discovery 4.6.1 Structural features and null models within ScienceWISE As we mentioned, the ScienceWISE system can be represented both as a bipartite (articlesconcepts) network and as its unipartite projection in the articles space. In the former case, links identify the presence of a given concept in a manuscript, while in the latter they represent the level of similarity between papers. The most striking difference between the two representations is related to their density: the number of links grows as O(N^2) in the unipartite case (where N is the number of articles in the collection), while the bipartite one leads to much sparser networks, whose number of links grows linearly with the number of articles. 4.6.2 Scientific activity dynamics within ScienceWISE Moreover, we have also added a further layer to the aforementioned representation, namely the similarity given by the reference lists of the articles (bibliographic coupling): we have therefore built a unipartite two- layers network. By performing a hierarchical clustering procedure, we have shown that the presence of this second layer significantly improves the quality of the outcome. In particular, we observe a much better partition at the bottom of the hierarchy, meaning that the presence of citations between papers allows us to identify the very specific research topics. 4.6.3 Temporal Analysis of Scientific Discoveries within ScienceWISE In the figure below, we consider the distribution of points for the concepts extracted from all the articles submitted during the year 2013 to arXiv.org in physics subject classes as principal category. The vast majority of hand- labeled generic concepts (black dots) sit along the diagonal, suggesting that generic concepts tend to maximize their entropy and, consequently, carry little information. However, a more careful analysis hints at the presence of many more concepts whose observed entropy is close to the maximum

Inner composition of the communities obtained for the 2013 dataset (physics manuscripts submitted to arXiv.org during that year). The color of each cell accounts for the fraction of articles of a given category belonging to a cluster (each column sums to 1). The articles of the same categories tend to incorporate into single clusters as justified by clearly visible block-diagonal structure of both idf and bp partitions. Nevertheless, the split of some categories into distinct clusters may be observed. For instance, the articles of nucl-th category are roughly equally split among hep- and cond-mat dominated categories. On the right, the most representative concepts for each cluster are shown. Specifically, we make use of Shannon’s entropy to quantify the amount of information carried out by concepts. For each concept, we compute two different entropies: one associated to the observed term- frequency (TF) probability distribution, and the other associated to a probability distribution derived from the maximum entropy principle and constrained by the first moment and log-moment of the observed TF distribution. The maximum entropy value acts as an upper bound for the observed entropy, allowing determining if the observed value is big or small with respect to a quantity computed for the same concept. Hence, the comparison of the two entropy values constitutes the core of the methodology we used to filter the similarity network. POTENTIAL IMPACT DISSEMINATION ACTIVITY

As regards the Dissemination we organised • A web site for the Proactive initiative (http://www.dym-cs.eu), • A session at the Conference ICALP 2013 in Riga Latvia (Conference site and workshop site) • The conference “Horizons in Social Science” June 2013 Lucca Italy (Link) • School at École de Physique Les Houches 2014 (Link) • Conference ECCS’14 in Lucca (Link) • Satellite at ECCS’14 (Link) • Book Networks of Networks Springer Series (G. D’Agostino, A. Scala eds) • Conference NETSCI 15 Zaragoza (Link) • Satellite at NETSCI 15 Zaragoza (Link) • Satellite at CCS’15 Phoenix USA (Link) • Book Interconnected Networkshttp://www.springer.com/la/book/9783319239453 Springer Series (A. Garas eds) 2015 • Book Data Science and Complex Networks, OUP G. Caldarelli A. Chessa 2016 • Satellite to STATPHYS Conference Paris, France July 2016 (Link) • Workshop in Complex Networks Lipari September 2016 (Link) • Satellite to CCS’16 Amsterdam (Link) • 419 papers published

4.2 Use and dissemination of foreground

A plan for use and dissemination of foreground (including socio-economic impact and target groups for the results of the research) shall be established at the end of the project. It should, where appropriate, be an update of the initial plan in Annex I for use and dissemination of foreground and be consistent with the report on societal implications on the use and dissemination of foreground (section 4.3 – H). The plan should consist of:

§ Section A

This section should describe the dissemination measures, including any scientific publications relating to foreground. Its content will be made available in the public domain thus demonstrating the added-value and positive impact of the project on the European Union.

§ Section B

This section should specify the exploitable foreground and provide the plans for exploitation. All these data can be public or confidential; the report must clearly mark non-publishable (confidential) parts that will be treated as such by the Commission. Information under Section B that is not marked as confidential will be made available in the public domain thus demonstrating the added-value and positive impact of the project on the European Union. Section A (public)

This section includes two templates

§ Template A1: List of all scientific (peer reviewed) publications relating to the foreground of the project.

§ Template A2: List of all dissemination activities (publications, conferences, workshops, web sites/applications, press releases, flyers, articles published in the popular press, videos, media briefings, presentations, exhibitions, thesis, interviews, films, TV clips, posters).

These tables are cumulative, which means that they should always show all publications and activities from the beginning until after the end of the project. Updates are possible at any time.

template A1: list of scientific (peer reviewed) publications, starting with the most important ones Number, date Permanent Is/ or frequency identifiers[1] Wil l op en Pla Year acc ce Title of the of ess NO Publi of Relevan Title Main author periodical or the publi . sher pub t pages [2] series VOL ISSU catio lica (if available) pro UME E n tion vid ed to this pu blic

ati on?

Universal resilience Gao, Jianxi and Barzel, 10.1038/nature169 1 patterns in complex Baruch and Barabási, Nature 530 7590 2016 307-312 yes 48 networks Albert-László Scholtes, Ingo and Causality-driven Wider, Nicolas and slow-down and Pfitzner, René and Nature 10.1038/ncomms60 2 speed-up of diffusion 5 2014 5024 yes Garas, Antonios and communications 24 in non-markovian Tessone, Claudio J and temporal networks Schweitzer, Frank Compensating for population sampling Génois, Mathieu and in simulations of Vestergaard, Christian L Nature 10.1038/ncomms98 3 6 2015 8860 yes epidemic spread on and Cattuto, Ciro and Communications 60 temporal contact Barrat, Alain networks Majdandzic, Antonio and Braunstein, Lidia A Multiple tipping and Curme, Chester and points and optimal Nature 10.1038/ncomms10 4 Vodenska, Irena and 7 2016 10850 yes repairing in communications 850 Levy-Carciente, Sary interacting networks and Eugene Stanley H and Havlin, Shlomo Orsini, Chiara and Dankulov, Marija M. Quantifying and Colomer-de-Simón, Nature 10.1038/ncomms96 5 randomness in real Pol and Jamakovic, 6 2015 8627 yes Communications 27 networks Almerima and Mahadevan, Priya and Vahdat, Amin and

Bassler, Kevin E. and Toroczkai, Zoltán and Boguñá Marián and Caldarelli, Guido and Fortunato, Santo and Krioukov, Dmitri Chiara Orsini, Marija Mitrović Dankulov, Almerima Jamakovic, Priya Mahadevan, Pol Colomer-de-Simón, Quantifying Amin Vahdat, Kevin E. Nature 10.1038/ncomms96 6 randomness in real 6 2015 8627 yes Bassler, Zoltán Communications 27 networks Toroczkai, Marián Boguñá, Guido Caldarelli, Santo Fortunato, Dmitri Krioukov Zhao, Jichang and Li, Spatio-temporal Daqing and Sanhedrai, propagation of Nature 10.1038/ncomms10 7 Hillel and Cohen, 7 2016 10094 yes cascading overload Communications 094 Reuven and Havlin, failures Shlomo Dorogovtsev, S. N. and 8 Ranking scientists Nature Physics 11 11 2015 882-883 10.1038/nphys3533 yes Mendes, J. F. F. Yan, Gang and Spectrum of Tsekenis, Georgios and controlling and Barzel, Baruch and 9 Nature Physics 11 9 2015 779-786 10.1038/nphys3422 yes observing complex Slotine, Jean-jacques networks and Liu, Yang-yu and Barabási, Albert-László Spontaneous Majdandzic, Antonio 10 Nature Physics 10 1 2013 34-38 10.1038/nphys2819 yes recovery in and Podobnik, Boris and

dynamical networks Buldyrev, Sergey V and Kenett, Dror Y and Havlin, Shlomo and Stanley, H Eugene The extreme Bashan, Amir and vulnerability of Berezin, Yehiel and 11 interdependent Nature Physics 9 10 2013 667-672 10.1038/nphys2727 yes Buldyrev, Sergey V SV spatially embedded and Havlin, Shlomo networks Quantifying long- Wang, Dashun and 10.1126/science.12 12 term scientific Song, Chaoming and Science 342 6154 2013 127-32 yes 37825 impact Barabási, Albert-László Control of coupled oscillator networks Skardal, Per Sebastian e150033 10.1126/sciadv.150 13 with application to Science Advances 1 7 2015 yes and Arenas, Alex 9 0339 microgrid technologies Poncela-Casasnovas, Julia and Gutiérrez- Roig, Mario and Gracia- Humans display a Lázaro, Carlos and reduced set of Vicens, Julian and e160045 10.1126/sciadv.160 14 consistent behavioral Science Advances 2 8 2016 yes Gómez-Gardeñes, Jesús 1 0451 phenotypes in dyadic and Perelló Josep and games Moreno, Yamir and Duch, Jordi and Sánchez, Angel P. Klimek, M. Systemic trade risk e150052 10.1126/sciadv.150 15 Obersteiner, and S. Science Advances 1 10 2015 yes of critical resources 2 0522 Thurner The dynamic of J. Borge-Holthoefer, N. information-driven Perra, B. Goncçalves, S. coordination e150115 10.1126/sciadv.150 16 González-Bailón, A. Science Advances 2 4 2016 yes phenomena: a 8 1158 Arenas, Y. Moreno, and transfer entropy A. Vespignani analysis Abruptness of Pahwa, Sakshi and 17 Scientific Reports 4 2014 10.1038/srep03694 yes cascade failures in Scoglio, Caterina and power grids Scala, Antonio Activity clocks: Gauvin, Laetitia and spreading dynamics Panisson, André and 18 on temporal Scientific Reports 3 2013 3099 10.1038/srep03099 yes Cattuto, Ciro and Barrat, networks of human Alain contact Assortativity and leadership emerge I. Sendiña-Nadal, M. M. from anti-preferential Danziger, Z. Wang, S. 19 Scientific Reports 6 2016 21297 10.1038/srep21297 yes attachment in Havlin, and S. heterogeneous Boccaletti networks Biased imitation in coupled evolutionary Santos, Md and 20 games in Dorogovtsev, Sn and Scientific Reports 1 2014 gen-19 10.1038/srep04436 yes interdependent Mendes, Jff networks Characterising two- C. Poletto, S. Meloni, A. pathogen V. Metre, V. Colizza, Y. 21 competition in Scientific Reports 5 2015 7895 10.1038/srep07895 yes Moreno, and A. spatially structured Vespignani environments Piškorec, Matija and Antulov-Fantulin, Nino Cohesiveness in and Novak, Petra Kralj financial news and 22 and Mozetič Igor and Scientific Reports 4 Vix 2014 5038 10.1038/srep05038 yes its relation to market Grčar, Miha and volatility Vodenska, Irena and Smuc, Tomislav Concurrent enhancement of Y.-H. Eom, S. 23 percolation and Boccaletti, and G. Scientific Reports 6 2016 27111 10.1038/srep27111 yes synchronization in Caldarelli adaptive networks Credit default swaps Puliga, Michelangelo 24 networks and and Caldarelli, Guido Scientific Reports 4 2014 6822 10.1038/srep06822 yes systemic risk and Battiston, Stefano Detection of gene Cantini, Laura and communities in Medico, Enzo and 25 multi-networks Scientific Reports 5 2015 17386 10.1038/srep17386 yes Fortunato, Santo and reveals cancer Caselle, Michele drivers Cardillo, Alessio and Gómez-Gardeñes, Jesús and Zanin, Emergence of Massimiliano and 26 network features Scientific Reports 3 2013 1344 10.1038/srep01344 yes Romance, Miguel and from multiplexity Papo, David and del Pozo, Francisco and Boccaletti, Stefano Epidemic risk from friendship network data: an equivalence Fournet, Julie and 27 Scientific Reports 6 2016 24593 10.1038/srep24593 yes with a non-uniform Barrat, Alain sampling of contact networks Fingerprinting Gauvin, Laetitia and temporal networks of Panisson, André and 28 Scientific Reports 3 2013 261-266 10.1038/srep03099 yes close-range human Cattuto, Ciro and Barrat, proximity Alain and Cattuto, Ciro Ramos, Marlon and Shao, Jia and Reis, How does public Saulo D S and 29 opinion become Anteneodo, Celia and Scientific Reports 5 2015 10032 10.1038/srep10032 yes extreme? Andrade, José S and Havlin, Shlomo and Makse, Hernán a Wang, Zhen and Xia, Impact of social Cheng-Yi and Meloni, punishment on 30 Sandro and Zhou, Scientific Reports 3 2013 3055 10.1038/srep03055 yes cooperative behavior Chang-Song and in complex networks Moreno, Yamir Islanding the power Mureddu, Mario and 31 grid on the Caldarelli, Guido and Scientific Reports 6 2016 34797 10.1038/srep34797 yes transmission level: Damiano, Alfonso and less connections for Scala, Antonio and more security Meyer-Ortmanns, Hildegard Berezin, Yehiel and Localized attacks on Bashan, Amir and spatially embedded 32 Danziger, Michael M Scientific Reports 5 2015 8934 10.1038/srep08934 yes networks with and Li, Daqing and dependencies Havlin, Shlomo Multiplexity versus correlation: the role Gemmetto, V. and 33 Scientific Reports 5 2015 9120 10.1038/srep09120 yes of local constraints in Garlaschelli, D. real multiplexes Cerina, Federica and Network Chessa, Alessandro and 34 communities within Scientific Reports 4 2014 4546 10.1038/srep04546 yes Pammolli, Fabio and and across borders Riccaboni, Massimo Vivaldo, Gianna and Networks of plants: Masi, Elisa and how to measure 35 Pandolfi, Camilla and Scientific Reports 2016 01-nov 10.1038/srep27077 yes similarity in Mancuso, Stefano and vegetable species Caldarelli, Guido Opinion dynamics on Quattrociocchi, Walter, interacting networks: 36 Guido Caldarelli, and Scientific Reports 4 2014 4938 10.1038/srep04938 yes media competition Antonio Scala and social influence Muchnik, Lev and Pei, Sen and Parra, Lucas C Origins of power-law and Reis, Saulo D S and degree distribution in Havlin, Shlomo and 37 the heterogeneity of Scientific Reports 3 2013 gen-23 10.1038/srep01783 yes Andrade, José S and human activity in Makse, Hernán A and social networks Havlin, Shlomo and Makse, Hernán A Physiologically motivated multiplex Sadilek, Maximilian and 38 Kuramoto model Scientific Reports 5 2015 10015 10.1038/srep10015 yes Thurner, Stefan describes phase diagram of cortical activity Di Muro M. A. and La Recovery of Rocca C. E. and 39 interdependent Stanley, H. E. and Scientific Reports 6 7600 2015 22834 10.1038/srep22834 yes networks Havlin, S. and Braunstein, L. A. Cuesta, Jose a and Reputation drives Gracia-la, Carlos and cooperative Gracia-Lázaro, Carlos 40 behaviour and Scientific Reports 5 C 2015 01-giu 10.1038/srep07843 yes and Ferrer, Alfredo and network formation in Moreno, Yamir and human groups Sánchez, Angel Strategical incoherence Matamalas, Joan T and regulates cooperation Poncela-Casasnovas, 41 Scientific Reports 5 2015 9519 10.1038/srep09519 yes in social dilemmas Julia and Gómez, Sergio on multiplex and Arenas, Alex networks Systemic risk Cimini, Giulio and analysis on Squartini, Tiziano and 42 reconstructed Scientific Reports 5 2015 15758 10.1038/srep15758 yes Garlaschelli, Diego and economic and Gabrielli, Andrea financial networks The scaling of human contacts and M. Tizzoni, K. Sun, D. 43 epidemic processes Benusiglio, M. Karsai, Scientific Reports 5 2015 15111 10.1038/srep15111 yes in metapopulation and N. Perra networks Time varying Karsai, Márton and networks and the 44 Perra, Nicola and Scientific Reports 4 2014 4001 10.1038/srep04001 yes weakness of strong Vespignani, Alessandro ties Proceedings of Collective credit Shen, Hua-Wei and the National 10.1073/pnas.1401 45 2014 01-giu yes allocation in science Barabási, Albert-László Academy of 992111 Science Impact of Gavaldà Arnau and Proceedings of 15322- 10.1073/pnas.1309 46 111 43 2014 yes heterogeneity and Choffnes, David and the National 15327 389111 socio-economic Otto, John S and Academy of factors in massive Sanchez, Mario A and Science decentralized sharing Bustamante, Fabian E ecosystems and Amaral, a N and Duch, Jordi and Guimerà Roger Navigability of De Domenico Manlio Proceedings of interconnected and Solé-Ribalta, Albert the National 10.1073/pnas.1318 47 111 23 2014 8351-6 yes networks under and Gómez, Sergio and Academy of 469111 random failures Arenas, Alex Science Proceedings of The price of S. Battiston, G. the National 10031- 10.1073/pnas.1521 48 complexity in Caldarelli, R. May, T. 110 2016 yes Academy of 10036 573113 financial networks Roukny, and J. Stiglitz Science Baxter, G. J. and Avalanche collapse Dorogovtsev, S. N. and Physical Review 10.1103/PhysRevL 49 of interdependent 109 24 2012 248701 yes Goltsev, A. V. and Letters ett.109.248701 networks Mendes, J. F. F. Betweenness Pfitzner, René and preference: Scholtes, Ingo and quantifying Physical Review 10.1103/PhysRevL 50 Garas, Antonios and 110 19 2013 198701 yes correlations in the Letters ett.110.198701 Tessone, Claudio J. and topological dynamics Schweitzer, Frank of temporal networks Breaking of Squartini, Tiziano and ensemble De Mol Joey and Den Physical Review 10.1103/PhysRevL 51 115 26 2015 01-mag yes equivalence in Hollander Frank and Letters ett.115.268701 networks Garlaschelli, Diego Moinet, Antoine and Burstiness and aging Starnini, Michele and Physical Review 10.1103/PhysRevL 52 in social temporal 114 2014 108701 yes Pastor-Satorras, Letters ett.114.108701 networks Romualdo Ruan, Zhongyuan and Kinetics of social Iñiguez, Gerardo and Physical Review 10.1103/PhysRevL 53 115 21 2015 gen-19 yes contagion Karsai, Márton and Letters ett.115.218702 Kertész, János Modeling temporal Barrat, Alain and Physical Review 10.1103/PhysRevL 54 110 15 2013 158702 yes networks using Fernandez, Bastien and Letters ett.110.158702 random itineraries Lin, Kevin K. and Young, Lai-Sang Zhou, Dong and Teleconnection paths Gozolchiani, Avi and Physical Review 10.1103/PhysRevL 55 via climate network 115 26 2015 01-mag yes Ashkenazy, Yosef and Letters ett.115.268501 direct link detection Havlin, Shlomo Skarpalezos, Loukas and Kittas, Aristotelis Anomalous biased Physical Review 10.1103/PhysRevE. 56 and Argyrakis, Panos 88 2013 12817 yes diffusion in networks E 88.012817 and Cohen, Reuven and Havlin, Shlomo Granell, Clara and Benchmark model to Darst, Richard K. and assess community Physical Review 10.1103/PhysRevE. 57 Arenas, Alex and 92 1 2015 12805 yes structure in evolving E 92.012805 Fortunato, Santo and networks Gómez, Sergio Tremblay, Nicolas and Bootstrapping under Barrat, Alain and Forest, constraint for the Cary and Nornberg, Physical Review 10.1103/PhysRevE. 58 assessment of group 88 5 2013 52812 yes Mark and Pinton, Jean- E 88.052812 behavior in human Francois and Borgnat, contact networks Pierre Collective frequency Skardal, Per Sebastian variation in network and Taylor, Dane and Physical Review 10.1103/PhysRevE. 59 93 4 2016 42314 yes synchronization and Sun, Jie and Arenas, E 93.042314 reverse PageRank Alex Communicability reveals a transition to Estrada, Ernesto and Physical Review 10.1103/PhysRevE. 60 coordinated behavior 89 4 2014 01-giu yes Gómez-Gardeñes, Jesús E 89.042819 in multiplex networks Cozzo, Emanuele and Contact-based social Baños, Raquel A. and Physical Review 10.1103/PhysRevE. 61 contagion in 88 5 2013 50801 yes Meloni, Sandro and E 88.050801 multiplex networks Moreno, Yamir Core organization of Azimi-Tafreshi, N, Physical Review 10.1103/PhysRevE. 62 directed complex Dorogovtsev, SN, 2012 yes E 87.032815 networks Mendes, JFF Baxter, Gareth J. and Correlated edge Bianconi, Ginestra and Physical Review 10.1103/PhysRevE. 63 overlaps in multiplex Da Costa Rui A. and 94 1 2016 12303 yes E 94.012303 networks Dorogovtsev, Sergey N. and Mendes, José F F Critical behavior of the relaxation rate, Yoon, S. and Sorbaro the susceptibility, Sindaci M. and Goltsev, Physical Review 10.1103/PhysRevE. 64 and a pair correlation 91 3 2015 32814 yes a. V. and Mendes, J. F. E 91.032814 function in the F. kuramoto model on scale-free networks Critical behaviour of De Nigris Sarah and the xy -rotors model Leoncini, Xavier and Physical Review 10.1103/PhysRevE. 65 88 2013 12131 yes on regular and small Nigris, Sarah De and E 88.012131 world networks Leoncini, Xavier Da Costa R. A. and Critical exponents of Dorogovtsev, S. N. and Physical Review 10.1103/PhysRevE. 66 the explosive 89 4 2014 01-giu yes Goltsev, A. V. and E 89.042148 percolation transition Mendes, J. F F Lee, K.-E. E. and Lopes, M. A. and Mendes, J. F. F. and Goltsev, A. V. Critical phenomena and K.-E. Lee, M. A. and noise-induced Physical Review 10.1103/PhysRevE. 67 Lopes, J. F. F. Mendes 1 2014 12701 yes phase transitions in E 89.012701 and A. V. Goltsev and neuronal networks Lee, K.-E. E. and Lopes, M. A. and Mendes, J. F. F. and Goltsev, A. V. Shai, Saray and Kenett, Critical tipping point Dror Y. and Kenett, distinguishing two Yoed N. and Faust, Physical Review 10.1103/PhysRevE. 68 types of transitions in 92 6 2015 62805 yes Miriam and Dobson, E 92.062805 modular network Simon and Havlin, structures Shlomo Effect of the Wang, Huijuan and Li, Physical Review 10.1103/PhysRevE. 69 interconnected Qian and D'Agostino, 88 2 2013 22801 yes E 88.022801 network structure on Gregorio and Havlin, the epidemic Shlomo and Stanley, H. threshold Eugene and Van Mieghem Piet and D'Agostino, Gregorio and Havlin, Shlomo and Stanley, H. Eugene and Van Mieghem Piet Efficiency of message transmission using L. Skarpalezos, A. Physical Review 10.1103/PhysRevE. 70 biased random walks Kittas, P. Argyrakis, R. 91 1 2015 12817 yes E 91.012817 in complex networks Cohen, and S. Havlin in the presence of traps Enhancing the R. Sevilla-Escoboza, R. stability of the Gutiérrez, G. Huerta- Physical Review 10.1103/PhysRevE. 71 synchronization of Cuellar, S. Boccaletti, J. 92 3 2015 32804 yes E 92.032804 multivariable Gómez-Gardeñes, A. coupled oscillators Arenas, and J. M. Buldú Estimating topological Cimini, Giulio and properties of Squartini, Tiziano and Physical Review 10.1103/PhysRevE. 72 92 4 2015 40802 yes weighted networks Gabrielli, Andrea and E 92.040802 from limited Garlaschelli, Diego information Cardillo, Alessio and Evolutionary Petri, Giovanni and dynamics of time- Nicosia, Vincenzo and Physical Review 10.1103/PhysRevE. 73 2013 52825 yes resolved social Sinatra, Roberta and E 90.052825 interactions Gómez-Gardeñes, Jesús and Latora, Vito Generalized achlioptas process Giazitzidis, Paraskevas Physical Review 10.1103/PhysRevE. 74 for the delay of 2013 24801 yes and Argyrakis, Panos E 88.024801 criticality in the percolation process Generalized Gutiérrez, R. and Physical Review 10.1103/PhysRevE. 75 88 5 2013 52908 yes synchronization in Sevilla-Escoboza, R. E 88.052908 relay systems with and Piedrahita, P. and instantaneous Finke, C. and Feudel, U. coupling and Buldú J. M. and Huerta-Cuellar, G. and Jaimes-Reátegui, R. and Moreno, Y. and Boccaletti, S. Giant components in Azimi-Tafreshi, N. and Physical Review 10.1103/PhysRevE. 76 directed multiplex Dorogovtsev, S N and 90 5 2014 01-lug yes E 90.052809 networks Mendes, J F F Hierarchical mutual information for the comparison of J. I. Perotti, C. J. Physical Review 10.1103/PhysRevE. 77 hierarchical Tessone, and G. 92 6 2015 62825 yes E 92.062825 community Caldarelli structures in complex networks How breadth of degree distribution influences network X. Yuan, S. Shao, H. E. Physical Review 10.1103/PhysRevE. 78 92 3 2015 32122 yes robustness: Stanley, and S. Havlin E 92.032122 comparing localized and random attacks How damage R. Burkholz, A. Garas, Physical Review 10.1103/PhysRevE. 79 diversification can 93 4 2016 42313 yes and F. Schweitzer E 93.042313 reduce systemic risk How memory generates Vestergaard, Christian Physical Review 10.1103/PhysRevE. 80 heterogeneous L. and Génois, Mathieu 90 4 2014 42805 yes E 90.042805 dynamics in temporal and Barrat, Alain networks Hybrid phase transition into an D. Lee, S. Choi, M. Physical Review 10.1103/PhysRevE. 81 absorbing state: Stippinger, J. Kertész, 93 4 2016 42109 yes E 93.042109 percolation and and B. Kahng avalanches Hyperbolicity M. Borassi, A. Chessa, Physical Review 10.1103/PhysRevE. 82 92 2015 32812 yes measures and G. Caldarelli E 92.032812 “democracy” in real- world networks Improving the performance of Darst, Richard K. and Physical Review 10.1103/PhysRevE. 83 algorithms to find Nussinov, Zohar and 2013 01-ago yes E 89.032809 communities in Fortunato, Santo networks R. A. Da Costa, S. N. Inverting the Dorogovtsev, A. V. Physical Review 10.1103/PhysRevE. 84 achlioptas rule for 91 4 2015 42130 yes Goltsev, and J. F. F. E 91.042130 explosive percolation Mendes K-core percolation Azimi-Tafreshi, N. and Physical Review 10.1103/PhysRevE. 85 on multiplex Gómez-Gardeñes, J. and 90 3 2014 01-set yes E 90.032816 networks Dorogovtsev, S. N. Kuramoto model Coutinho, B. C. and with frequency- Goltsev, A. V. and Physical Review 10.1103/PhysRevE. 86 87 3 2013 32106 yes degree correlations Dorogovtsev, S. N. and E 87.032106 on complex networks Mendes, J. F. F. Message passing theory for Cellai, D. and percolation models Physical Review 10.1103/PhysRevE. 87 Dorogovtsev, S.N. and 94 2016 32301 yes on multiplex E 94.032301 Bianconi, G. networks with link overlap Murase, Yohsuke and Multi-layer weighted Physical Review 10.1103/PhysRevE. 88 Jo, Hang-hyun and 90 2014 52810 yes social network model E 90.052810 Kaski, Kimmo Multiple percolation transitions in a Bianconi, Ginestra and Physical Review 10.1103/PhysRevE. 89 configuration model 89 6 2014 01-ott yes Dorogovtsev, Sergey N. E 89.062814 of a network of networks V. Gemmetto, T. Multiplexity and Squartini, F. Picciolo, F. Physical Review 10.1103/PhysRevE. 90 multireciprocity in 94 4 2016 42316 yes Ruzzenenti, and D. E 94.042316 directed multiplexes Garlaschelli Mutually connected Bianconi, Ginestra and Physical Review 10.1103/PhysRevE. 91 91 1 2014 01-set yes component of Dorogovtsev, Sergey N. E 91.012804 network of networks and Mendes, José Fernando F. Y. H. Eom, A. Perna, S. Network-based Fortunato, E. Darrouzet, Physical Review 10.1103/PhysRevE. 92 model of the growth 92 6 2015 62810 yes G. Theraulaz, and C. E 92.062810 of termite nests Jost Optimal transport G. Li, Reis, S. D. S., Physical Review 10.1103/PhysRevE. 93 exponent in spatially Moreira, A. A., and 87 2013 42810 yes E 87.042810 embedded networks Havlin, S Gao, Jianxi and Percolation of a Buldyrev, Sergey V. and Physical Review 10.1103/PhysRevE. 94 general network of Stanley, H. Eugene and 88 6 2013 gen-13 yes E 88.062816 networks Xu, Xiaoming and Havlin, Shlomo Hu, Yanqing and Zhou, Percolation of Dong and Zhang, Rui interdependent Physical Review 10.1103/PhysRevE. 95 and Han, Zhangang and 88 5 2013 01-lug yes networks with E 88.052805 Rozenblat, Céline and intersimilarity Havlin, Shlomo Percolation of Zhou, Di and Gao, partially Jianxi and Stanley, H. Physical Review 10.1103/PhysRevE. 96 87 5 2013 52812 yes interdependent scale- Eugene and Havlin, E 87.052812 free networks Shlomo Preferential Foschi, Rachele and attachment in Schiavo, Stefano and Physical Review 10.1103/PhysRevE. 97 90 2 2014 22817 yes multiple trade Riccaboni, Massimo and E 90.022817 networks Schiavo, Stefano Bastas, Nikolaos and Maragakis, Michalis and Argyrakis, Panos and Random walk with Ben-Avraham, Daniel priorities in Physical Review 10.1103/PhysRevE. 98 and Havlin, Shlomo and 88 2 2013 22803 yes communicationlike E 88.022803 Carmi, Shai and Ben- networks Avraham, Daniel and Havlin, Shlomo and Carmi, Shai Reaction-diffusion Physical Review 10.1103/PhysRevE. 99 Garas, Antonios 92 2 2015 20801 yes processes on E 92.020801 interconnected scale- free networks Reconstructing the Mastrandrea, Rossana world trade and Squartini, Tiziano Physical Review 10.1103/PhysRevE. 100 multiplex: the role of 90 6 2014 gen-18 yes and Fagiolo, Giorgio E 90.062804 intensive and and Garlaschelli, Diego extensive biases Robustness of a Shekhtman, Louis M. network formed of and Berezin, Yehiel and Physical Review 10.1103/PhysRevE. 101 90 1 2014 01-lug yes spatially embedded Danziger, Michael M. E 90.012809 networks and Havlin, Shlomo Dong, Gaogao and Gao, Robustness of Jianxi and Du, Ruijin Physical Review 10.1103/PhysRevE. 102 network of networks and Tian, Lixin and 87 2013 52804 yes E 87.052804 under targeted attack Stanley, H. Eugene and Havlin, Shlomo G. Timar, S. N. Scale-free networks Physical Review 10.1103/PhysRevE. 103 Dorogovtsev, and J. F. 94 2 2016 22302 yes with exponent one E 94.022302 F. Mendes Simultaneous first- Zhou, Dong and and second-order Bashan, Amir and percolation Cohen, Reuven and Physical Review 10.1103/PhysRevE. 104 90 1 2014 01-lug yes transitions in Berezin, Yehiel and E 90.012803 interdependent Shnerb, Nadav and networks Havlin, Shlomo Solution of the explosive percolation Da Costa R. A. and quest. II. Infinite- Dorogovtsev, S. N. and Physical Review 10.1103/PhysRevE. 105 order transition 91 3 2015 gen-15 yes Goltsev, A. V. and E 91.032140 produced by the Mendes, J. F. F. initial distributions of clusters Structural and functional properties Emmerich, Thorsten and Physical Review 10.1103/PhysRevE. 106 of spatially Bunde, Armin and 89 6 2014 01-lug yes E 89.062806 embedded scale-free Havlin, Shlomo networks 107 Synchronization in M. A. Lopes, E. M. Physical Review 94 1 2016 12308 10.1103/PhysRevE. yes the random field Lopes, S. Yoon, J. F. F. E 94.012308 Kuramoto model on Mendes, and A. V. complex networks Goltsev Temporal percolation Starnini, Michele and Physical Review 10.1103/PhysRevE. 108 in activity driven 89 2013 ott-13 yes Satorras, Rp E 89.032807 networks Value of peripheral Zhang, Yan and Garas, nodes in controlling Physical Review 10.1103/PhysRevE. 109 Antonios and 93 1 2016 12309 yes multilayer scale-free E 93.012309 Schweitzer, Frank networks Baxter, Gareth J. and Weak percolation on Dorogovtsev, Sergey N. Physical Review 10.1103/PhysRevE. 110 89 4 2014 gen-14 yes multiplex networks and Mendes, José F F E 89.042801 and Cellai, Davide Weighted networks Caldarelli, Guido and as randomly Chessa, Alessandro and Physical Review 10.1103/PhysRevE. 111 87 2 2013 20106 yes reinforced urn Crimaldi, Irene and E 87.020106 processes Pammolli, Fabio Community MacMahon, Mel and Physical Review 10.1103/PhysRevX 112 detection for 5 2 2015 gen-34 yes Garlaschelli, Diego X .5.021006 correlation matrices Conditions for viral influence spreading Hu, Yanqing and Physical Review 10.1103/PhysRevX 113 through multiplex Havlin, Shlomo and 4 2 2014 yes X .4.021031 correlated social Makse, Hernán a. networks Baxter, G. J. and Critical dynamics of Dorogovtsev, S. N. and Physical Review 10.1103/PhysRevX 114 the k-core pruning Lee, K. E. and Mendes, 5 3 2015 31017 yes X .5.031017 process J. F F and Goltsev, A. V. Double percolation Colomer-de-Simon, Pol phase transition in Physical Review 4102 10.1103/PhysRevX 115 and Boguñá 1 2014 41020 yes clustered complex X 0 .4.041020 Mariániversitat De networks Sanz, Joaquín and Xia, Dynamics of Physical Review 10.1103/PhysRevX 116 Cheng-Yi and Meloni, 4 2014 41005 yes interacting diseases X .4.041005 Sandro and Moreno, Yamir Effects of network structure, Gleeson, James P. and competition and O'Sullivan, Kevin P. and Physical Review 10.1103/PhysRevX 117 6 2 2016 21019 yes memory time on Baños, Raquel A. and X .6.021019 social spreading Moreno, Yamir phenomena Hidden connectivity in networks with Physical Review 10.1103/PhysRevX 118 Krause, Sebastian M. 6 4 2016 41022 yes vulnerable classes of X .6.041022 nodes Identifying modular flows on multilayer networks reveals M. De Domenico, A. Physical Review 10.1103/PhysRevX 119 highly overlapping Lancichinetti, A. 5 1 2015 11027 yes X .5.011027 organization in Arenas, and M. Rosvall interconnected systems Identifying modular De Domenico Manlio flows on multilayer and Lancichinetti, networks reveals Physical Review 1102 10.1103/PhysRevX 120 Andrea and Arenas, 2014 gen-14 yes highly overlapping X 7 .5.011027 Alex and Rosvall, organization in social Martin systems Network structure, metadata and the D. Hric, T. P. Peixoto, Physical Review 10.1103/PhysRevX 121 6 3 2016 31038 yes prediction of missing and S. Fortunato X .6.031038 nodes Unbiased community MacMahon, Mel and Physical Review 10.1103/PhysRevX 122 detection for 5 2013 gen-22 yes Garlaschelli, Diego X .5.021006 correlation matrices A network inference method for large- PLoS scale unsupervised Guimerà Roger and e100337 10.1371/journal.pc 123 Computational 9 12 2013 yes identification of Sales-Pardo, Marta 4 bi.1003374 Biology novel drug-drug interactions Host mobility drives Poletto, Chiara and pathogen Meloni, Sandro and PLoS e100316 10.1371/journal.pc 124 competition in Colizza, Vittoria and Computational 9 8 2013 yes 9 bi.1003169 spatially structured Moreno, Yamir and Biology populations Vespignani, Alessandro How to estimate PLoS epidemic risk from Mastrandrea, Rossana 10.1371/journal.pc 125 Computational 12 6 2016 gen-19 yes incomplete contact and Barrat, Alain bi.1005002 Biology diaries data? Klimek, Peter and Quantification of Kautzky-Willer, diabetes comorbidity PLoS Alexandra and Chmiel, 10.1371/journal.pc 126 risks across life using Computational 11 4 2015 gen-16 yes Anna and Schiller- bi.1004125 nation-wide big Biology Frühwirth, Irmgard and claims data Thurner, Stefan Caldarelli, Guido and A multi-level Chessa, Alessandro and geographical study of Pammolli, Fabio and 10.1371/journal.po 127 Italian political Pompa, Gabriele and PLoS ONE 9 5 2014 e95809 yes ne.0095809 elections from twitter Puliga,Michelangelo data and Riccaboni, Massimo and Riotta, Gianni Competing for attention in social L. Feng, Y. Hu, B. Li, e012609 10.1371/journal.po 128 media under H. E. Stanley, S. Havlin, PLoS ONE 10 7 2015 yes 0 ne.0126090 information overload and L. A. Braunstein conditions Contact patterns Fournet, Julie and 10.1371/journal.po 129 among high school PLoS ONE 9 9 2014 e107878 yes Barrat, Alain ne.0107878 students Contact patterns in a high school: a comparison between Mastrandrea, Rossana e013649 10.1371/journal.po 130 data collected using and Fournet, Julie and PLoS ONE 10 9 2015 yes 7 ne.0136497 wearable sensors, Barrat, Alain contact diaries and friendship surveys 131 Coupling news Ranco, Gabriele and PLoS ONE 11 1 2016 e014657 10.1371/journal.po yes sentiment with web Bordino, Ilaria and 6 ne.0146576 browsing data Bormetti, Giacomo and improves prediction Caldarelli, Guido and of intra-day price Lillo, Fabrizio and dynamics Treccani, Michele DebtRank: a Bardoscia, Marco and microscopic Battiston, Stefano and e013040 10.1371/journal.po 132 PLoS ONE 10 6 2015 yes foundation for shock Caccioli, Fabio and 6 ne.0130406 propagation Caldarelli, Guido Detecting the community structure and activity patterns Gauvin, Laetitia and of temporal 10.1371/journal.po 133 Panisson, André and PLoS ONE 9 1 2014 yes networks: a non- ne.0086028 Cattuto, Ciro negative tensor factorization approach Distress propagation M. Bardoscia, F. in complex networks: Caccioli, J. I. Perotti, G. e016382 10.1371/journal.po 134 PLoS ONE 11 10 2015 yes the case of non-linear Vivaldo, and G. 5 ne.0163825 DebtRank Caldarelli Zollo, Fabiana and Novak, Petra Kralj and Del Vicario Michela and Emotional dynamics Bessi, Alessandro and e013874 10.1371/journal.po 135 in the age of PLoS ONE 10 9 2015 yes Mozetič Igor and Scala, 0 ne.0138740 misinformation Antonio and Caldarelli, Guido and Quattrociocchi, Walter Vanhems, Philippe and Barrat, Alain and Estimating potential Cattuto, Ciro and infection Pinton, J.-F. and transmission routes 10.1371/journal.po 136 Khanafer, Nagham and PLoS ONE 8 9 2013 gen-17 yes in hospital wards ne.0073970 Régis, C. and Kim, using wearable Byeul-a and Comte, proximity sensors Brigitte and Voirin, Nicolas and Pinton, J.-F. and Khanafer, Nagham and Régis, C. and Kim, Byeul-a and Comte, Brigitte and Voirin, Nicolas Popović Marko and štefančić Hrvoje and Extraction of Sluban, Borut and temporal networks Novak, Petra Kralj and 10.1371/journal.po 137 from term co- PLoS ONE 9 2014 e99515 yes Grčar, Miha and ne.0099515 occurrences in online Mozetič Igor and Puliga, textual sources Michelangelo and Zlatić Vinko B. Lengyel, A. Varga, Geographies of an e013724 10.1371/journal.po 138 B. Sagvari, A. Jakobi, PLoS ONE 10 9 2015 yes online social network 8 ne.0137248 and J. Kertesz Green power grids: Mureddu, Mario and how energy from Caldarelli, Guido and e013531 10.1371/journal.po 139 renewable sources Chessa, Alessandro and PLoS ONE 10 9 2015 yes 2 ne.0135312 affects networks and Scala, Antonio and markets Damiano, Alfonso Squartini, Tiziano and Information recovery Ser-Giacomi, Enrico e012507 10.1371/journal.po 140 in behavioral PLoS ONE 10 5 2015 yes and Garlaschelli, Diego 7 ne.0125077 networks and Judge, George Long-range Kitsak, Maksim and correlations and Elmokashfi, Ahmed and 10.1371/journal.po 141 memory in the PLoS ONE 10 11 2015 01-dic yes Havlin, Shlomo and ne.0181 dynamics of internet Krioukov, Dmitri interdomain routing Mesoscopic Almog, Assaf and community structure Besamusca, Ferry and 10.1371/journal.po 142 of financial markets PLoS ONE 10 7 2015 gen-16 yes MacMahon, Mel and ne.0133679 revealed by price and Garlaschelli, Diego sign fluctuations Modeling the role of Murase, Yohsuke and e013300 10.1371/journal.po 143 relationship fading Jo, Hang-Hyun and PLoS ONE 10 7 2015 yes 5 ne.0133005 and breakup in social Török, János and network formation Kertész, János and Kaski, Kimmo Multilingual twitter sentiment Mozetič Igor and Grčar, e015503 10.1371/journal.po 144 classification: the Miha and Smailović PLoS ONE 11 5 2016 yes 6 ne.0155036 role of human Jasmina annotators Science vs Bessi, Alessandro and conspiracy : Coletto, Mauro and e011809 10.1371/journal.po 145 collective narratives Davidescu, George PLoS ONE 10 2015 yes 3 ne.0118093 in the age of Alexandru and Scala, misinformation Antonio Science vs conspiracy: A. Bessi, M. Coletto, G. e011809 10.1371/journal.po 146 collective narratives A. Davidescu, and A. PLoS ONE 10 2 2015 yes 3 ne.0118093 in the age of Scala misinformation Self-healing Quattrociocchi Walter, networks: 10.1371/journal.po 147 Caldarelli Guido, Scala PLoS ONE 9 2 2014 e87986 yes redundancy and ne.0087986 Antonio structure Kralj Novak Petra and Smailovic, Jasmina and e014429 10.1371/journal.po 148 Sentiment of emojis PLoS ONE 10 12 2015 yes Sluban, Borut and 6 ne.0144296 Mozetic, Igor Ranco, Gabriele and The effects of twitter Aleksovski, Darko and e013844 10.1371/journal.po 149 sentiment on stock Caldarelli, Guido and PLoS ONE 10 9 2015 yes 1 ne.0138441 price returns Grčar, Miha and Mozetič Igor The rise of china in Zhu, Zhen, Federica the international Cerina, Alessandro 10.1371/journal.po 150 trade network: a Chessa, Guido PLoS ONE 9 8 2014 e105496 yes ne.0105496 community core Caldarelli, and Massimo detection approach Riccaboni The role of the Sáenz-Royo, Carlos and e012607 10.1371/journal.po 151 organization Gracia-Lázaro, Carlos PLoS ONE 10 5 2015 yes 6 ne.0126076 structure in the and Moreno, Yamir diffusion of innovations Bessi, Alessandro and Zollo, Fabiana and Del Trend of narratives Vicario Michela and e013464 10.1371/journal.po 152 in the age of PLoS ONE 10 8 2015 yes Scala, Antonio and 1 ne.0134641 misinformation Caldarelli, Guido and Quattrociocchi, Walter Twitter-based Eom, Young-Ho and analysis of the Puliga, Michelangelo e013118 10.1371/journal.po 153 dynamics of and Smailović Jasmina PLoS ONE 10 7 2015 yes 4 ne.0131184 collective attention to and Mozetič Igor and political parties Caldarelli, Guido Bessi, Alessandro and Zollo, Fabiana and Del Vicario Michela and Users polarization on Puliga, Michelangelo e015964 10.1371/journal.po 154 Facebook and PLoS ONE 11 8 2016 yes and Scala, Antonio and 1 ne.0159641 Youtube Caldarelli, Guido and Uzzi, Brian and Quattrociocchi, Walter Approximate pure Nash equilibria in Caragiannis, Ioannis and ACM weighted congestion Fanelli, Angelo and Transactions on 155 games: existence, 3 1 2015 2 10.1145/2614687 yes Gravin, Nick and Economics and efficient Skopalik, Alexander Computation computation, and structure Colini-Baldeschi, On multiple keyword ACM Riccardo and Henzinger, sponsored search Transactions on PAR 10.1007/978-3- 156 Monika and Leonardi, 4 2015 01-dic yes auctions with Economics and T 2 642-31585-5-1 Stefano and Starnberger, budgets Computation Martin Anagnostopoulos, Aris Stochastic query ACM and Becchetti, Luca and covering for fast Transactions on 11:1- 157 Bordino, Ilaria and 33 3 2015 10.1145/2699671 yes approximate Information 11:35 Leonardi, Stefano and document retrieval Systems. Mele, Ida and Sankowski, Piotr On existence and Algorithmic properties of Game Theory: M. Drees, M. Feldotto, approximate pure Proceedings of 10.1017/CBO9780 158 S. Riechers, and A. 2015 178-189 yes Nash equilibria in the 8th 511800481 Skopalik bandwith allocation International games Symposium Algorithmic Game Theory: Designing cost- G. Christodoulou, S. Proceedings of 10.1007/978-3- 159 sharing methods for Leonardi, and A. 2016 yes the 9th 662-53354-3_26 Bayesian games Sgouritsa International Symposium Anagnostopoulos, Aris Online network and Grandoni, Fabrizio 10.1007/s00453- 160 Algorithmica 76 1 2016 88-109 yes design with outliers and Leonardi, Stefano 015-0021-y and Sankowski, Piotr Annals of the Inter-cities' Japanese multinational firms 10.1111/glob.1213 161 C. Rozenblat Association of 61 2015 219-237 yes networks and 0 Economic gravitation model Geographers Retweet networks of the European Cherepnalkoski, Darko Applied Network 10.1007/s41109- 162 parliament: 1 2016 2 yes and Mozetič Igor Science 016-0001-4 evaluation of the community structure Approximation Shortest augmenting B. Bosek, D. Leniowski, and Online 10.1007/978-3- 163 paths for online P. Sankowski, and A. Algorithms: 13th 2015 59-71 yes 319-28684-6 matchings on trees Zych International Workshop Jo, Hang-Hyun and A static model for Murase, Yohsuke and 164 stylized facts in Török, János and ArXiv 2016 arXiv:1611.03664 yes social networks Kertész, János and Kaski, Kimmo An algebraic Simas, Tiago, Mario topological method Chavez, Pablo 165 for multimodal brain ArXiv 2015 arXiv:1504.02265 yes Rodriguez, and Albert networks Diaz-guilera comparisons Analysis of financial Kralj Novak P and 166 news with ArXiv 2015 arXiv:1508.00027 yes Grčar, M and Mozetič I Newsstream P. Balister, B. Bollobás, Coalescence on the 167 J. Lee, and B. ArXiv 2016 arXiv:1610.07430 yes real line Narayanan Congestion games M. Feldotto, L. Leder, 168 with mixed ArXiv 2016 arXiv:1607.05143 yes and A. Skopalik objectives Consistent random P. Balister, B. Bollobás, 169 vertex-orderings of ArXiv 2015 arXiv:1506.03343 yes and S. Janson graphs Counting connected Bolloas, Bela, and 170 hypergraphs via the ArXiv 2014 arXiv:1404.5887 yes Riordan Oliver probabilistic method Counting dense connected B. Bollobas and O. 171 ArXiv 2015 arXiv:1511.04739 yes hypergraphs via the Riordan probabilistic method Cycles and clustering G. J. Baxter, D. Cellai, 172 in multiplex S. N. Dorogovtsev, and ArXiv 2016 arXiv:1609.05788 yes networks J. F. F. Mendes Akrida, Eleni C, Designing and Gasieniek Leszek, 173 testing temporally ArXiv 2015 arXiv:1502.04579 yes George B Mertzios, and connected graphs Paul G Spirakis Darst, Richard K. and Detection of Granell, Clara and timescales in Arenas, Alex and 174 ArXiv 2016 arXiv:1604.00758 yes evolving complex Gómez, Sergio and systems Saramäki, Jari and Fortunato, Santo Early prediction of movie box office Mestyan, Martin and 175 success based on Taha, Yasseri and ArXiv 2012 arXiv:1211.0970 yes wikipedia activity Kertész, János big data Essential Balister, Paul, Bela 176 enhancements Bollobas, Oliver ArXiv 2014 arXiv:1402.0834 yes revisited Riordan Bessi, Alessandro and Zollo, Fabiana and Everyday the same Vicario, Michela Del 177 picture: popularity and Scala, Antonio and ArXiv 2015 arXiv:1501.07201 yes and content diversity Petroni, Fabio and Gonçalves, Bruno and Quattrociocchi, Walter Bollobás, Béla and Exploring Riordan, Oliver and 178 hypergraphs with ArXiv 2014 arXiv:1403.6558 yes Bollobás, Béla and martingales Riordan, Oliver Find your place: L. Becchetti, A. simple distributed Clementi, E. Natale, F. 179 ArXiv 2016 arXiv:1511.03927 yes algorithms for Pasquale, and L. community detection Trevisan Zhu, Zhen, Michelangelo Puliga, 180 Global value trees Federica Cerina, ArXiv 2014 arXiv:1410.4694 yes Alessandro Chessa, and Massimo Riccaboni. Graph minors for preserving terminal Y. K. Cheung, G. distances 181 Goranci, and M. ArXiv 2016 arXiv:1604.08342 yes approximately - Henzinger lower and upper bounds Quattrociocchi, Walter Influence of media 182 and Caldarelli, Guido ArXiv 2013 arXiv:1307.4292 yes on collective debates and Scala, Antonio 183 Lifespan method as a Mata, S and Boguna, ArXiv 2014 arXiv:1411.5835 yes tool to study Marian and Castellano, criticality in Claudio and Pastor- absorbing-state phase Satorras, Romualdo transitions Balister, Paul N. and Bollobás, Béla and Lee, 184 Line percolation ArXiv 2014 arXiv:1403.6851 yes Jonathan D. and Narayanan, Bhargav P. Mapping the G. Timár, A. V. structure of directed Goltsev, S. N. 185 ArXiv 2016 arXiv:1607.0069 yes networks: beyond the Dorogovtsev, and J. F. "bow-tie" diagram F. Mendes Metastable Ferreira, R. S. and da localization of Costa, R. A. and 186 ArXiv 2016 arXiv:1606.05624 yes diseases in complex Dorogovtsev, S. N. and networks Mendes, J. F. F. Mining concurrent Panisson, a, L Gauvin, 187 topical activity in M Quaggiotto, and C ArXiv 2014 arXiv:1403.1403 yes microblog streams Cattuto Model reproduces Starnini, Michele and individual, group and Baronchelli, Andrea and 188 collective dynamics ArXiv 2014 arXiv:1409.0507 yes Pastor-satorras, of human contact Romualdo networks Modeling peer and M. PiSkorec, N. external influence in Antulov-Fantulin, I. 189 ArXiv 2016 arXiv:1610.08262 yes online social Miholic, T. Smuc, and networks M. Sikic Multilayer stochastic Valles-Catala, Toni and block models reveal Massucci, Francesco 190 the multilayer and Guimera, Roger and ArXiv 2013 arXiv:1411.1098 yes structure of complex Sales-Pardo, Marta and networks Vall, Toni J. Kertesz, J. Torok, Y. Multiplex modeling 191 Murase, H.-H. Jo, and ArXiv 2016 arXiv:1609.08381 yes of the society K. Kaski Non-consensus Li, Q, Braunstein, LA, 192 ArXiv arXiv:1210.0862 yes opinion models on Wang, H, Shao, J, complex networks Stanley, HE, Havlin, S Nucleation and B. Bollobás, S. 193 growth in two Griffiths, R. Morris, L. ArXiv 2015 arXiv:1508.06267 yes dimensions Rolla, and P. Smith On the maximum B. Bollobás, M. running time in 194 Przykucki, O. Riordan, ArXiv 2015 arXiv:1510.07096 yes graph bootstrap and J. Sahasrabudhe percolation Optimal redundancy Krause, Sebastian M against disjoint 195 and Danziger, Michael ArXiv 2015 arXiv:1503.04058 yes vulnerabilities in M and Zlatić Vinko networks Percolation in networks of networks Bianconi, Ginestra and 196 with random ArXiv 2014 arXiv:1411.4160 yes Dorogovtsev, Sergey N matching of nodes in different layers Becchetti, L and Clementi, Andrea and Plurality consensus 197 Natale, E and Pasquale, ArXiv 2014 arXiv:1407.2565 yes in the gossip model Francesco and Silvestri, Riccardo Random geometric P. Balister, B. Bollobás, graphs and 198 K. Gunderson, I. ArXiv 2015 arXiv:1504.05324 yes isometries of normed Leader, and M. Walters spaces Regulation of García-Pérez, Guillermo burstiness by 199 and Boguñá Marián and ArXiv 2014 arXiv:1410.3816 yes network-driven Serrano, María ángeles activation Revealing latent factors of temporal Gauvin, Laetitia and networks for Panisson, André and 200 ArXiv 2015 arXiv:1501.02758 yes mesoscale Barrat, Alain and intervention in Cattuto, Ciro epidemic spread 201 Revenue maximizing Colini-Baldeschi, ArXiv 2016 arXiv:1610.09221 yes envy-free pricing in Riccardo and Leonardi, matching markets Stefano and Zhang, with budgets Qiang Sousa da Mata, Slow relaxation Angelica, Romualdo dynamics and aging Pastor-Satorras, 202 in random walks on ArXiv 2014 arXiv:1411.4601 yes Angelica Mata, and activity driven Romualdo Pastor- temporal networks Satorras Slow-down vs. Scholtes, I, Wider, N, Speed-up of Pfitzner, R, Garas, A, 203 information diffusion ArXiv 2013 arXiv:1307.4030 yes Tessone, CJuan, in non-markovian Schweitzer, F temporal networks Subcritical u- Balister, Paul and bootstrap percolation Bollobás, Béla and 204 models have non- ArXiv 2013 arXiv:1311.5883 yes Przykucki, Michał and trivial phase Smith, Paul transitions Systemic risk management in M. V. Leduc, S. 205 financial networks ArXiv 2016 arXiv:1601.02156 yes Poledna, and S. Thurner with credit default swaps The double role of GDP in shaping the Almog Assaf, Squartini 206 structure of the Tiziano, and ArXiv 2015 arXiv:1512.02454 yes international trade Garlaschelli Diego network Li, Aming and The fundamental Cornelius, Sean P. and 207 advantages of Liu, Yang-Yu and ArXiv 2016 arXiv:1607.06168 yes temporal networks Wang, Long and Barabási, Albert-László Tomasello, Mario The rise and fall of Vincenzo and 208 ArXiv 2013 arXiv:1304.3623 yes r&d networks Napoletano, Mauro and Garas, Antonios and Schweitzer, Frank Bollobás, Béla and The sharp threshold Duminil-Copin, Hugo 209 ArXiv 2016 arXiv:1603.05237 yes for the Duarte model and Morris, Robert and Smith, Paul The threshold for Bollobás, Béla and 210 jigsaw percolation on Riordan, Oliver and ArXiv 2015 arXiv:1503.05186 yes random graphs Smith, Paul Bollobás, Béla and Universality of two- Duminil-Copin, Hugo 3175 211 dimensional critical ArXiv 2014 70 arXiv:1406.6680 yes and Morris, Robert and 32 cellular automata Smith, Paul Cerina, Federica and World input-output Zhu, Zhen and Chessa, 212 ArXiv 2014 24 arXiv:1407.0225 yes network Alessandro and Riccaboni, Massimo An infectious disease model on empirical Machens, Anna and networks of human Gesualdo, Francesco contact: bridging the and Rizzo, Caterina and BMC infectious 10.1186/1471- 213 13 2013 185 yes gap between Tozzi, Alberto E and diseases 2334-13-185 dynamic network Barrat, Alain and data and contact Cattuto, Ciro matrices Contact diaries versus wearable proximity sensors in T. Smieszek, S. Castell, measuring contact BMC infectious 10.1186/s12879- 214 A. Barrat, C. Cattuto, P. 16 2016 341 yes patterns at a diseases 016-1676-y J. White, and G. Krause conference: method comparison and participants' attitudes Mitigation of infectious disease at Gemmetto, Valerio and BMC Infectious 10.1186/s12879- 215 school: targeted class Barrat, Alain and 14 1 2014 01-ott yes Diseases 014-0695-9 closure vs school Cattuto, Ciro closure Enhancing the evaluation of pathogen Mastrandrea, Rossana transmission risk in a and Soto-Aladro, BMC research 10.1186/s13104- 216 hospital by merging Alberto and Brouqui, 8 2015 426 yes notes 015-1409-0 hand-hygiene Philippe and Barrat, compliance and Alain contact data: a proof- of-concept study On controlling Skardal, Per Sebastian 217 networks of limit- Chaos 26 9 2016 10.1063/1.4954273 yes and Arenas, Alex cycle oscillators Gracia-Lazaro, Carlos Cooperation in and Floría, Luis M. LM changing and Gómez-Gardeñes, environments: Jesús and Moreno, Chaos, Solitons 10.1016/j.chaos.20 218 irreversibility in the Yamir and Gracia- 56 2013 188-193 yes and Fractals 13.08.008 transition to Lázaro, Carlos and cooperation in Floría, Luis M. LM and complex networks Gómez-Gardeñes, Jesús and Moreno, Yamir Recent advances on Shekhtman, Louis M. failure and recovery Chaos, Solitons 10.1016/j.chaos.20 219 and Danziger, Michael 90 2015 28-36 yes in networks of and Fractals 16.02.002 M. and Havlin, Shlomo networks Measuring contact patterns with wearable sensors: Barrat, A. and Cattuto, methods, data C. and Tozzi, AE. and Clin. Microbiol. 10.1111/1469- 220 20 2014 10 yes characteristics and Vanhems, P. and Voirin, Infect. 0691.12472 applications to data- N. driven simulations of infectious diseases COCOON Locality-sensitive A. Pacuk, P. Sankowski, Springer 10.1007/s00453- 221 hashing without false K. Wegrzycki, and P. 2016 105-118 yes International 016-0176-1 negatives for lp Wygocki Publishing 222 Sentiment leaning of B. Sluban, J. Smailovic, Computational 2 1 2015 gen-21 10.1186/s40649- yes influential S. Battiston, and I. Social Networks 015-0016-5 communities in Mozetic social networks Mocanu, Delia and Collective attention Rossi, Luca and Zhang, Computers in 10.1016/j.chb.2015 223 in the age of Qian and Karsai, Màrton 0 2015 yes Human Behavior .01.024 (mis)information Marton and Quattrociocchi, Walter Directional depletion Scala, A. De Sanctis Condensed Matter 10.5488/CMP.17.3 224 interactions in 17 3 2014 33007 yes Lucentini P. Physics 3007 shaped particles Critical The complexity Information Fioriti, Vincenzo and science approach vs. Infrastructure 10.4018/978-1- 225 D'Agostino, Gregorio 2013 139-152 yes the simulative Protection and 4666-2964-6.ch007 and Scala, Antonio approach Resilience in the ICT Sector Scala, Antonio and D Critical 'agostino Gregorio and The robustness of Information 10.1007/978-3- 226 D'Agostino, Gregorio 6983 2013 223-226 yes assortativity Infrastructure 642-41476-3_22 and D 'agostino Security Gregorio D. Tuckett, A. Mandel, Uncertainty, decision D. Mangalagiu, A. science, and policy Abramson, J. Hinkel, K. 10.1080/08913811. 227 Critical Review 3811 2015 gen-30 yes making: a manifesto Katsikopoulos, A. 2015.1037078 for a research agenda Kirman, T. Malleret, and I. Mozetic Tour A. Anagnostopoulos, R. Data Mining and 10.1007/s10618- 228 recommendation for Atassi, L. Becchetti, A. Knowledge 2016 gen-32 yes 016-0477-7 groups Fazzone, and F. Silvestri Discovery Decision and Game Theory for Network elicitation M. Dziubinski, P. Security: 10.1007/978-3- 229 in adversarial Sankowski, and Q. Proceedings of 2016 397-414 yes 642-17197-0 environment Zhang the 7th International Conference Interset: interactive M. Mihelcic and T. 10.1007/978-3- 230 redescription set Discover Science 9956 2016 35-50 yes Smuc 540-88411-8 exploration Simple and efficient local codes for Michail, Othon and Distributed 10.1007/s00446- 231 29 3 2016 207-237 yes distributed stable Spirakis, Paul G. Computing 015-0257-4 network construction Bootstrap percolation Gunderson, Karen and Electronic Journal 10.1214/EJP.v19- 232 on Galton–Watson 19 13 2013 yes Holmgren, Cecilia of Probability 2758 trees Gao, Jianxi and Liu, Recent progress on Xueming and Li, 12187- 10.3390/en8101218 233 the resilience of Energies 8 10 2015 yes Daqing and Havlin, 12210 7 complex networks Shlomo Baños, Raquel and Borge-Holthoefer, Diffusion dynamics Javier and Wang, Ning 4553- 234 with changing Entropy 15 11 2013 10.3390/e15114553 yes and Moreno, Yamir and 4568 network composition González-Bailón, Sandra Low-temperature Garlaschelli, Diego and behaviour of social Ahnert, Sebastian and 3148- 235 Entropy 15 8 2013 10.3390/e15083238 yes and economic Fink, Thomas and 3169 networks Caldarelli, Guido Ground truth? Concept-based Palchykov, Vasyl and communities versus Gemmetto, Valerio and 10.1140/epjds/s136 236 EPJ Data Science 5 1 2016 yes the external Boyarsky, Alexey and 88-016-0090-4 classification of Garlaschelli, Diego physics manuscripts De Domenico Manlio Personalized routing and Lima, Antonio and 10.1140/epjds/s136 237 for multitudes in EPJ Data Science 4 1 2015 01-nov yes González, Marta C. and 88-015-0038-0 smart cities Arenas, Alex Predicting human Beiró M. G. and mobility through the Panisson, A. and 10.1140/epjds/s136 238 EPJ Data Science 2016 17 yes assimilation of social Tizzoni, M. and Cattuto, 88-016-0092-2 media traces into C. mobility models Sarigöl, Emre and Predicting scientific Pfitzner, Rene and success based on 10.1140/epjds/s136 239 Scholtes, Ingo and EPJ Data Science 3 9 2014 21 yes coauthorship 88-014-0009-x Garas, Antonios and networks Schweitzer, Frank Kiti, Moses C. and Tizzoni, Michele and Kinyanjui, Timothy M. Quantifying social and Koech, Dorothy C. contacts in a and Munywoki, Patrick household setting of 10.1140/epjds/s136 240 K. and Meriac, Milosch EPJ Data Science 5 1 2016 yes rural kenya using 88-016-0084-2 and Cappa, Luca and wearable proximity Panisson, André and sensors Barrat, Alain and Cattuto, Ciro and Nokes, D. James Alvarez, Raquel and Sentiment cascades Garcia, David and 10.1140/epjds/s136 241 in the 15m EPJ Data Science 4 1 2015 gen-13 yes Moreno, Yamir and 88-015-0042-4 movement Schweitzer, Frank Auletta, Vincenzo and Logit dynamics with Ferraioli, Diodato and ESA Lecture concurrent updates 10.1007/978-3- 242 Pasquale, Francesco and Notes in 8125 2013 73-84 yes for local interaction 642-40450-4_7 Penna, Paolo and Computer Science games Persiano, Giuseppe Clementi, Andrea E F and Crescenzi, Pierluigi and Doerr, Carola and Rumor spreading in ESA Lecture Fraigniaud, Pierre and 10.1007/978-3- 243 random evolving Notes in 8125 2013 325-336 yes Isopi, Marco and 642-40450-4_28 graphs Computer Science Panconesi, Alessandro and Pasquale, Francesco and Silvestri, Riccardo Impact of spatially Vestergaard, Christian L European Journal constrained sampling and Valdano, Eugenio 10.1017/S0956792 244 of Applied 2016 gen-17 yes of temporal contact and Génois, Mathieu 516000309 Mathematics networks on the and Poletto, Chiara and evaluation of the Colizza, Vittoria and epidemic risk Barrat, Alain Intersections of Bollobás, Béla and European Journal 10.1016/j.ejc.2014. 245 random hypergraphs 44 2015 125-139 yes Scott, Alex of Combinatorics 08.023 and tournaments Positive independence densities of finite Balister, Paul and European Journal 10.1016/j.ejc.2016. 246 rank countable Bollobás, Béla and 57 2016 85-93 yes of Combinatorics 04.005 hypergraphs are Gunderson, Karen achieved by finite hypergraphs Higher-order aggregate networks Scholtes, Ingo and European in the analysis of 10.1140/epjb/e2016 247 Wider, Nicolas and Physical Journal 89 3 2016 gen-15 yes temporal networks: -60663-0 Garas, Antonios B path structures and centralities Network of participants in European Tsouchnika, Maria and 10.1140/epjb/e2014 248 european research: Physical Journal 87 12 2014 292 yes Argyrakis, Panos -50450-4 accepted versus B rejected proposals Dynamic instability of cooperation due to Xia, Cheng-Yi and diverse activity Meloni, Sandro and Europhysics 5800 10.1209/0295- 249 2015 06-nov yes patterns in Perc, Matjaz and Letters 2 5075/109/58002 evolutionary social Moreno, Yamir dilemmas Zhang, Xin and Shao, Dynamic motifs in Shuai and Stanley, H. Europhysics 10.1209/0295- 250 socio-economic 108 5 2014 58001 yes Eugene and Havlin, Letters 5075/108/58001 networks Shlomo P. Piedrahíta, Borge- Modeling self- Holthoefer, J., Moreno, sustained activity Europhysics 10.1209/0295- 251 Y., and Arenas A. and 104 2013 48004 yes cascades in socio- Letters 5075/104/48004 Piedrahita, P and Borge- technical networks Holthoefer, J and Moreno, Yamir and Arenas, Alex Percolation F. Wang, D. Li, X. Xu, Europhysics 10.1209/0295- 252 properties in a traffic 112 3 2015 38001 yes R. Wu, and S. Havlin Letters 5075/112/38001 model Static and dynamic Bastas, N. and behavior of multiplex Lazaridis, F. and Europhysics 10.1209/0295- 253 networks under 109 3 2015 38006 yes Argyrakis, P. and Letters 5075/109/38006 interlink strength Maragakis, M. variation Huang, Xuqing and Shao, Shuai and Wang, The robustness of Huijuan and Buldyrev, Europhysics 10.1209/0295- 254 interdependent 101 1 2013 18002 yes Sergey V. and Eugene Letters 5075/101/18002 clustered networks Stanley H. and Havlin, Shlomo Mihelčić Matej and A framework for Džeroski, Sašo and Expert Systems 10.1016/j.eswa.201 255 redescription set 68 2017 196-215 yes Lavrač Nada and šmuc, with Applications 6.10.012 construction Tomislav Transference for the J. Balogh, B. Bollobás, Forum Math. 10.1017/fms.2015. 256 erdħøs-ko-rado 3 2015 18 yes and B. P. Narayanan Sigma 21 theorem The multipolar regionalization of C. Rozenblat, F. Zaidi, 10.1111/glob.1213 257 world cities in the Global Networks 2016 yes and A. Bellwald 0 multinational firms networks Handbook of Baxter, G.J and Optimization in Heterogeneous k- Dorogovtsev, S. N. and Complex 10.1007/978-1- 258 core organisation in 57 2012 229-252 yes Goltsev, A. V. and F., Networks : 4614-0754-6_9 complex networks Mendes J. F. Theory and Applications Bosek, Bartlomiej, Online bipartite Dariusz Leniowski, IEEE Computer 10.1109/FOCS.201 259 matching in offline 2014 yes Piotr Sankowski, and society 4.48 time Anna Zych IEEE INFOCOM Harks, T and Hoefer, M 2014 - IEEE Routing games with 10.1109/INFOCO 260 and Schewior, K and Conference on 24 4 2014 352-360 yes progressive filling M.2014.6847957 Skopalik, A Computer Communications An efficient and fair Richa, Andréa W and IEEE/ACM mac protocol robust Scheideler, Christian 10.1109/TNET.201 261 Transactions on 21 3 2013 760-771 yes to reactive and Schmid, Stefan and 2.2210241 Networks interference Zhang, Jin In Revised On verifying and Selected Papers of maintaining the 11th E. C. Akrida and P. G. 10.1007/978-3- 262 connectivity of International 9536 2015 142-154 yes Spirakis 319-28472-9_11 interval temporal Symposium on networks Algorithms for Sensor Systems Bessi, Alessandro, Social determinants In Social Guido Caldarelli, of content selection Informatics - 6th 10.1007/978-3- 263 Michela Del Vicario, 2014 yes in the age of International 319-13734-6_18 Antonio Scala, and (mis)information Conference Walter Quattrociocchi Voirin, Nicolas and Payet, Cécile and Barrat, Alain and Combining high- Cattuto, Ciro and resolution contact Khanafer, Nagham and data with virological Infection control Régis, Corinne and 10.1017/ice.2014.5 264 data to investigate and hospital 36 3 2015 254-60 yes Kim, Byeul-A and 3 influenza epidemiology Comte, Brigitte and transmission in a Casalegno, Jean- tertiary care hospital Sébastien and Lina, Bruno and Vanhems, Philippe Genève plus tournée C. Rozenblat, G. Métral, 265 vers l'international INSEE-Analyse 46 2015 yes and A. Bellwald que lyon Interconnected Wider, Nicolas and Interconnected 10.1007/978-3- 266 2016 149-164 yes networks Garas, Antonios and Networks 319-23947-7 Scholtes, Ingo and Schweitzer, Frank Vulnerability of M. M. Danziger, L. M. interdependent Shekhtman, A. Bashan, Interconnected 10.1007/978-3- 267 2016 79-99 yes networks and Y. Berezin, and S. networks 319-23947-7_5 networks of networks Havlin Cascade failures and International distributed Scala, A, S Pahwa, and 10.1504/IJCIS. 268 Journal of Critical 11 1 2015 27-35 yes generation in power C Scoglio 2015.067395 Infrastructures grids P. Balister, B. Bollobás, Israel Journal of 3175 995- 10.1007/s11856- 269 Subtended angles Z. Füredi, I. Leader, and 214 2016 yes Mathematics 32 1012 016-1370-1 M. Walters Misinformation in Bessi, A, M Coletto, G the loop: the 270 A Davidescu, A Scala, ITAIS – Genova, 2014 yes emergence of and W Quattrociocchi narratives in osn Levy-Carciente, Sary Dynamical and Kenett, Dror Y. and Journal of macroprudential 10.1016/j.jbankfin. 271 Avakian, Adam and Banking & 59 2015 164-181 yes stress testing using 2015.05.008 Stanley, H. Eugene and Finance network theory Havlin, Shlomo Borge-Holthoefer, Cascading behaviour Javier and Banos, Journal of 10.1093/comnet/cnt 272 in complex socio- Raquel A and Gonzalez- Complex 1 1 2013 mar-24 yes 006 technical networks Bailon, Sandra and Networks Moreno, Yamir Percolation and Danziger, Michael M. Journal of cascade dynamics of and Bashan, Amir and 10.1093/comnet/cn 273 Complex 2014 yes spatial networks with Berezin, Yehiel and u020 Networks partial dependency Havlin, Shlomo Stationarity, non- stationarity and Journal of Squartini, Tiziano and 10.1093/comnet/cn 274 early-warning signals Complex 2014 12 yes Garlaschelli, Diego u012 in economic Networks networks To bail-out or to bail- Klimek, Peter and Journal of 10.1016/j.jedc.2014 275 50 2015 144-154 yes in? Answers from an Poledna, Sebastian and Economic .08.020 agent-based model Doyne Farmer J. and Dynamics and Thurner, Stefan Control The multi-layer Poledna, Sebastian and network nature of Molina-Borboa, José systemic risk and its Luis and Martínez- Journal of 10.1016/j.jfs.2015. 276 20 2015 70-81 yes implications for the Jaramillo, Serafín and Financial Stability 08.001 costs of financial van der Leij, Marco and crises Thurner, Stefan Redescription mining augmented with Mihelčić Matej and Journal of random forest of Dzeroski, Saso and Intelligent 277 16 D 2015 333 yes multi-target Lavrac, Nada and šmuc, Information predictive clustering Tomislav Systems trees Ensemble nonequivalence in D. Garlaschelli and F. Journal of Physics 278 0 2016 gen-35 yes random graphs with D. Hollander A modular structure Journal of Data reliability in Sanz, Joaquín and Statistical 10.1088/1742- 279 complex directed Cozzo, Emanuele and Mechanics: 2013 2014 P12008 5468/2013/12/P120 yes networks Moreno, Yamir Theory and 08 Experiment Emergence of giant Journal of strongly connected F. Caravelli, M. Statistical 10.1088/1742- 280 components in Bardoscia, and F. Mechanics: 5 2016 53211 5468/2016/05/0532 yes continuum disk-spin Caccioli Theory and 11 percolation Experiment Emergence of giant strongly connected Journal of components in Caravelli, Francesco and Statistical 10.1088/1742- continuum disk-spin 281 Bardoscia, Marco and Mechanics: 5 2016 53211 5468/2016/05/0532 yes percolation Caccioli, Fabio Theory and 11 emergence of giant Experiment scc in continuum disk-spin percolation Close relationships: a Palchykov, Vasyl and Journal of 03- 10.1007/s10955- 282 151 2013 735-744 yes study of mobile Kertész, János and Statistical Physics apr 013-0705-0 communication Dunbar, Robin and records Kaski, Kimmo Borge-Holthoefer, Emergence of Javier and Meloni, influential spreaders Journal of 01- 10.1007/s10955- 283 Sandro and Gonçalves, 151 2013 383-393 yes in modified rumor Statistical Physics feb 012-0595-6 Bruno and Moreno, models Yamir Immunization Starnini, Michele and strategies for Machens, Anna and Journal of 10.1016/j.jtbi.2013. 284 epidemic processes Cattuto, Ciro and Barrat, Theoretical 337 2013 89-100 yes 07.004 in time-varying Alain and Pastor- Biology. contact networks satorras, Romualdo Bessi, Alessandro and The economy of Scala, Antonio and Journal of Trust 10.1186/s40493- 285 attention in the age Rossi, Luca and Zhang, 1 1 2014 12 yes Management 014-0012-y of (mis)information Qian and Quattrociocchi, Walter Abshoff, Sebastian and Brief announcement : Cord-Landwehr, a model for Lecture Notes in 286 Andreas and Jung, 8768 2014 294 yes multilevel network Computer Science Daniel and Skopalik, games Alexander Colini-Baldeschi, Combinatorial Riccardo, Stefano Lecture Notes in 10.1007/978-3- 287 auctions with Leonardi, Piotr 8877 2014 423-428 yes Computer Science 319-13129-0_18 budgets Sankowski, and Qiang Zhang Efficiency of truthful Adamczyk, Marek, Piotr and symmetric Lecture Notes in 10.1007/978-3- 288 Sankowski, and Qiang 8768 2014 13-24 yes mechanisms in one- Computer Science 662-44803-8_2 Zhang sided matching Limiting price CIgler, Ludek and discrimination when Dvorak, Wolfgang and Lecture Notes in 10.1007/978-3- 289 selling products with 8877 2014 44-57 yes Henzinger, Monika and Computer Science 319-13129-0_4 positive network Starnberger, Martin externalities Locality-sensitive Pacuk, Andrzej and Lecture Notes in 10.1007/s00453- 290 9797 2016 105-118 yes hashing without false Sankowski, Piotr and Computer Science 016-0176-1 negatives for lp Wegrzycki, Karol and Wygocki, Piotr Minimum linear Eikel, Martina, Christian Lecture Notes in 10.1007/978-3- 291 arrangement of Scheideler, and 8952 2014 yes Computer Science 319-18263-6_15 series-parallel graphs Alexander Setzer Brunetti, Sara and Minimum weight Cordasco, Gennaro and Lecture Notes in 10.1007/978-3- 292 dynamo and fast Gargano, Luisa and 2012 yes Computer Science 642-34611-8_26 opinion spreading Lodi, Elena and Quattrociocchi, Walter Multilayer clustering Gamberger, Dragan and Lecture Notes in 10.1007/978-3- 293 : a discovery 8877 2014 87-98 yes Mihelˇ, Matej Computer Science 319-11812-3_8 experiment Abshoff, Sebastian, Multilevel network Andreas Cord- Lecture Notes in 10.1007/978-3- 294 8877 2014 435-440 yes games Landwehr, Daniel Jung, Computer Science 319-13129-0_36 and Alexander Skopalik Foreback, Dianne, On stabilizing Andreas Koutsopoulos, Lecture Notes in 295 departures in overlay Mikhail Nesterenko, 8756 2014 48-62 yes Computer Science networks Christian Scheideler, and Thim Strothmann E. C. Akrida, L. On temporally Gąasieniec, G. B. Lecture Notes in 10.1007/978-3- 296 connected graphs of 9499 2015 84-96 yes Mertzios, and P. G. Computer Science 319-28684-6_8 small cost Spirakis On two-party communication Abshoff, S, Benter, M, Lecture Notes in 10.1007/978-3- 297 8304 2013 yes through dynamic Malatyali, M Computer Science 319-03850-6_2 networks Quality of service in Cord-landwehr, Andreas Lecture Notes in 3175 10.1007/978-3- 298 network creation 8877 2014 423-428 yes and Alexander, M Computer Science 32 319-13129-0_34 games Reconstructing G. Cimini, T. Squartini, topological N. Musmeci, M. Puliga, Lecture Notes in 10.1007/978-3- 299 properties of A. Gabrielli, D. 8852 2015 323-333 yes Computer Science 319-15168-7_41 complex networks Garlaschelli, S. using the fitness Battiston, and G. model Caldarelli Revenue maximizing Colini-Baldeschi, envy-free fixed-price Riccardo and Leonardi, Lecture Notes in 10.1007/978-3- 300 8877 2014 233-246 yes auctions with Stefano and Sankowski, Computer Science 319-13129-0_18 budgets? Piotr and Zhang, Qiang Robust : a crash- Eikel Martina, failure-resistant Lecture Notes in 10.1007/978-3- 301 Scheideler Christian, 8878 2014 yes distributed storage Computer Science 319-14472-6_8 Setzer Alexander system Synthetic sequence generator for Antulov-Fantulin, Nino recommender and Bosnjak, Matko and Lecture Notes in 10.1007/978-3- 302 systems – memory Zlatic, Vinko and Grcar, 8777 2014 yes Computer Science 319-11812-3_3 biased random walk Miha and Smuc, on a sequence Tomislav multilayer network Adamczyk, Marek and Borodin, Allan and Web and Internet Lecture Notes in 10.1007/978-3- 303 Ferraioli, Diodato and 9470 2015 gen-15 yes economics Computer Science 319-13129-0 de Keijzer, Bart and Leonardi, Stefano Welfare Leibniz S. Bhattacharya, W. maximization with International 10.4230/LIPIcs.ST 304 Dvořák, M. Henzinger, 2013 gen-13 yes friends-of-friends Proceedings in ACS.2015.90 and M. Starnberger network externalities Informatics Network science: a useful tool in Kenett, Dror Y. and 10.1007/s11299- 305 Mind and Society 14 2 2015 155-167 yes economics and Havlin, Shlomo 015-0167-y finance Data on face-to-face Genois, Mathieu and contacts in an office Vestergaard, Christian L building suggest a and Fournet, Julie and 10.1017/nws.2015. 306 Network Science 3 3 2015 326-347 yes low-cost vaccination Panisson, Andre and 10 strategy based on Bonmarin, Isabelle and community linkers Barrat, Alain Ca-re-chord: a churn Benter, Markus and 10.1109/NetSys.20 307 resistant self- Divband, Mohammad Network Systems 2013 27-34 yes 13.11 stabilizing chord and Kniesburges, overlay network Sebastian and Koutsopoulos, Andreas and Graffi, Kalman Avalanches in Baxter, G. J. and Networks of multiplex and Dorogovtsev, S.N. and Networks: The 10.1007/978-3- 308 2014 37-52 yes interdependent Goltsev, A.V. and Last Frontier of 319-03518-5_2 networks Mendes, J.F.F. Complexity Networks of Battiston, Stefano and Networks: The 10.1007/978-3- 309 Financial networks 2014 311-321 yes Caldarelli, Guido Last Frontier of 319-03518-5_14 Complexity Relating ensemble diversity and Sluban, Borut and 10.1016/j.neucom.2 310 performance: a study Neurocomputing 2015 yes Lavrač Nada 014.10.086 in class noise detection New Frontiers in Redescription mining M. Mihelcic, S. Mining Complex with multi-target 10.1007/978-3- 311 Dzeroski, N. Lavrac, Patterns: 4th 2016 125-143 yes predictive clustering 319-39315-5_9 and T. Smuc International trees Workshop A GDP-driven model for the binary and Almog, Assaf and New Journal of 10.1088/1367- 312 weighted structure of Squartini, Tiziano and 17 1 2015 13009 yes Physics 2630/17/1/013009 the international Garlaschelli, Diego trade network Binary versus non- binary information in real time series: Almog, Assaf and New Journal of 10.1088/1367- 313 16 2014 93015 yes empirical results and Garlaschelli, Diego Physics 2630/16/9/093015 maximum-entropy matrix models Complex networks Scala, A and Auconi, P for data-driven and Scazzocchio, M and New Journal of 10.1088/1367- 314 medicine: the case of Caldarelli, G and 16 11 2014 115017 yes Physics 2630/16/11/115017 class iii dentoskeletal Mcnamara, JA and disharmony Franchi, L 315 Dynamical origins of P. Klimek, M. New Journal of 18 8 2016 83045 10.1088/1367- yes the community Diakonova, V. Eguiluz, Physics 2630/18/8/083045 structure of multi- M. S. Miguel, and S. layer societies Thurner Enhanced Mastrandrea, Rossana reconstruction of and Squartini, Tiziano New Journal of 10.1088/1367- 316 weighted networks 16 4 2014 43022 yes and Fagiolo, Giorgio Physics 2630/16/4/043022 from strengths and and Garlaschelli, Diego degrees Interdependent Danziger, Michael M. resistor networks New Journal of 10.1088/1367- 317 and Bashan, Amir and 17 4 2015 43046 yes with process-based Physics 2630/17/4/043046 Havlin, Shlomo dependency Vajna, Szabolcs and Modelling bursty New Journal of 10.1088/1367- 318 Tóth, Bálint and 15 2013 103203 yes time series Physics 2630/15/10/103023 Kertész, János Y. Wang, A. Oceanic El-Niño Gozolchiani, Y. New Journal of 10.1088/1367- 319 wave dynamics and 18 3 2015 33021 yes Ashkenazy, and S. Physics 2630/18/3/033021 climate networks Havlin Shao, Shuai and Huang, Percolation of Xuqing and Stanley, H. New Journal of 10.1088/1367- 320 localized attack on 17 2 2015 23049 yes Eugene and Havlin, Physics 2630/17/2/023049 complex networks Shlomo Quantifying sudden C. Masoller, Y. Hong, changes in dynamical S. Ayad, F. Gustave, S. New Journal of 10.1088/1367- 321 17 2 2015 23068 yes systems using Barland, A. J. Pons, S. Physics 2630/17/2/023068 symbolic networks Gómez, and A. Arenas Resilience of networks formed of L. M. Shekhtman, S. New Journal of 10.1088/1367- 322 17 2 2015 123007 yes interdependent Shai, and S. Havlin Physics 2630/17/12/123007 modular networks Spreading of diseases Chmiel, Anna and through comorbidity New Journal of 1150 10.1088/1367- 323 Klimek, Peter and 2014 115013 yes networks across life Physics 13 2630/16/11/115013 Thurner, Stefan and gender Triadic closure Klimek, Peter and New Journal of 10.1088/1367- 324 dynamics drives 15 2013 63008 yes Thurner, Stefan Physics 2630/15/6/063008 scaling laws in social multiplex networks Unbiased Squartini, Tiziano and New Journal of 10.1088/1367- 325 sampling of network Mastrandrea, Rossana 17 2 2014 23052 yes Physics 2630/17/2/023052 ensembles and Garlaschelli, Diego Scala, Antonio, Guido Caldarelli, Alessandro Nonlinear Chessa, Alfonso Phenomena in Power grids, smart Damiano, Mario Complex 10.1007/978-94- 326 grids and complex 2014 yes Mureddu, Sakshi Systems: From 017-8704-8_8 networks Pahwa, Caterina Nano to Macro Scoglio, and Walter Scale Quattrociocchi User-based representation of time-resolved L. Alessandretti, M. 10.1098/rsos.16015 327 Open Science 3 2016 160156 yes multimodal public Karsai, and L. Gauvin 6 transportation networks On the impact of Arregui, Sergio and masking and Sanz, Joaquín and blocking hypotheses 328 Marinova, Dessislava PeerJ 4 2016 e1513 10.7717/peerj.1513 yes for measuring the and Martín, Carlos and efficacy of new Moreno, Yamir tuberculosis vaccines Periodica On a problem of Bollobas, Bela and 10.1007/s10998- 329 Mathematica 72 1 2016 23-28 yes Erdos and Moser Scott, Alex 015-0098-8 Hungarica Layer - layer J. Gomez-Gardenes, M. Phylosophical competition in De Domenico, G. Transactions of 2015011 10.1098/rsta.2015.0 330 373 2016 yes multiplex complex Gutiérrez, A. Arenas, the Royal Society 7 117 networks and S. Gómez A Enhancing resilience Stippinger, Marcell and 10.1016/j.physa.20 331 of interdependent Physica A 416 2013 481-487 yes Kertész, János 14.08.069 networks by healing Sgrignoli, Paolo and The relation between Metulini, Rodolfo and 10.1109/SITIS.201 332 global migration and Physica A 417 2015 245-260 yes Schiavo, Stefano and 3.92 trade networks Riccaboni, Massimo A. Scala, P. G. De Cascades in Sanctis Lucentini, G. 323- 10.1016/j.physd.20 333 interdependent flow Physica D 2016 35-39 yes Caldarelli, and G. 324 15.10.010 networks D'Agostino Erosion of synchronization: P. S. Skardal, D. Taylor, 323- 10.1016/j.physd.20 334 coupling Physica D 2016 40-48 yes J. Sun, and A. Arenas 324 15.10.015 heterogeneity and network structure Nonlinear dynamics A. Arenas and M. De 323- 10.1016/j.physd.20 335 on interconnected Physica D 2016 01-apr yes Domenico 324 16.03.016 networks Random walk A. Solé-Ribalta, M. De centrality in 323- 10.1016/j.physd.20 336 Domenico, S. Gómez, Physica D 2016 73-79 yes interconnected 324 16.01.002 and A. Arenas multilayer networks Systemic risk in multiplex networks R. Burkholz, M. V. 323- 10.1016/j.physd.20 337 with asymmetric Leduc, A. Garas, and F. Physica D 2016 64-72 yes 324 15.10.004 coupling and Schweitzer threshold feedback Community detection in 10.1016/j.physrep.2 338 S. Fortunato and D. Hric Physics Reports 659 2016 gen-44 yes networks: a user 016.09.002 guide Proceedings of An ensemble N. Wider, A. Garas, I. ECCS 2014: 10.1007/978-3- 339 perspective on multi- Scholtes, and F. European 2016 37-59 yes 319-23947-7 layer networks Schweitzer Conference on Complex Systems G. Christodoulou, M. Proceedings of Strategic contention Gairing, S. Nikoletseas, European 10.4230/LIPIcs.ES 340 resolution with 71 2016 gen-16 yes C. Raptopoulos, and P. Symposium on A.2016.30 limited feedback Spirakis Algorithms On competitive Proceedings of A. Maecker, M. algorithms for IEEE 10.1109/IPDPS.20 341 Malatyali, and F. M. A. 2016 700-709 yes approximations of International 16.91 D. Heide top-k-position Parallel and monitoring of Distributed distributed streams Processing Symposium Proceedings of the 10th Community International B. Sluban, J. Smailovic, sentiment on Conference on 10.1109/SITIS.201 342 M. Juric, I. Mozetic, and 2015 376-382 yes environmental topics Signal-Image 4.27 S. Battiston in social networks Technology and Internet-Based Systems Proceedings of the 10th Targeted interest- A. Anagnostopoulos, F. International 527— ISBN 978-1- 343 driven advertising in 2016 yes Petroni, and M. Sorella Conference on 530 57735-758-2 cities using twitter Web and Social Media Proceedings of the 11th International A retweet network D. Cerepnalkoski and I. Conference on 350— 10.1109/SITIS.201 344 analysis of the 2015 yes Mozetic Signal-Image 357 5.8 European parliament Technology & Internet-Based Systems Proceedings of Sequential posted the 11th M. Adamczyk, A. price mechanisms International 10.1007/978-3- 345 Borodin, D. Ferraioli, B. 9470 2015 gen-15 yes with correlated Conference on 319-13129-0 Keijzer, and S. Leonardi valuations Web and Internet Economics Proceedings of Bidding strategies for A. Anagnostopoulos, R. the 12th 3175 346 fantasy-sports Cavallo, S. Leonardi, Conference on 2016 yes 32 auctions and M. Sviridenko Web and Internet Economics Proceedings of Revenue maximizing the 13th ACM envy-free pricing in R. Colini-Baldeschi, S. 10.1145/2229012.2 347 Conference on 2016 yes matching markets Leonardi, and Q. Zhang 229052 Electronic with budgets Commerce Proceedings of Hskip+: a self- the 14th IEEE stabilizing overlay Feldotto, Matthias and International 10.1109/P2P.2014. 348 network for nodes Scheideler, Christian 2014 yes Conference on 6934300 with heterogeneous and Graffi, Kalman Peer to Peer bandwidths Computing Proceedings of Competitive the 14th influence in social Anagnostopoulos, Aris, International networks: 1767- 349 Diodato Ferraioli, and Conference on 2014 arXiv:1401.4770 yes convergence, 1768 Stefano Leonardi Autonomous submodularity and Agents and Multi- competition effects Agent Systems Proceedings of Learning a the 15th IEEE A. Anagnostopoulos and 10.1109/ICDM.201 350 macroscopic model International 2015 685-690 yes M. Sorella 5.126 of cultural dynamics Conference on Data Mining Proceedings of Detecting anomalies the 15th IEEE A. Sapienza, A. in time-varying International 10.1109/ICDMW.2 351 Panisson, J. Wu, L. 2016 516-523 yes networks using Conference on 015.128 Gauvin, and C. Cattuto tensor decomposition Data Mining Workshop Proceedings of the 18th International Symposium on Self-stabilizing R. Gmyr, J. Lefèvre, and 10.1007/978-3- 352 Stabilization, 2016 yes metric graphs C. Scheideler 319-49259-9_20 Safety and Security of Distributed Systems Proceedings of the 19th ACM Abshoff, S, Benter, M, SIGKDD Token dissemination Cord-landwehr, A, International 10.1007/978-3- 353 in geometric 2013 yes Malatyali, M, Meyer- Conference on 642-45346-5_3 dynamic networks auf-der-Heide, F Knowledge Discovery and Data Mining Proceedings of the 19th ACM Abrahao, Bruno and SIGKDD Trace complexity of Chierichetti, Flavio and International 10.1145/2487575.2 354 2013 491-499 yes network inference Kleinberg, Robert and Conference on 487664 Panconesi, Alessandro Knowledge Discovery and Data Mining Proceedings of the 1st Anomaly detection in International temporal graph data: A. Sapienza, A. Workshop on 355 an iterative tensor Panisson, J. Wu, L. 1425 2015 117-121 yes Advanced decomposition and Gauvin, and C. Cattuto Analytics and masking approach Learning on Temporal Data Proceedings of Centrality rankings A. Solé-Ribalta, M. De the 2014 ACM 10.1145/2615569.2 356 in multiplex Domenico, S. Gómez, 2014 149-155 yes Conference on 615687 networks and A. Arenas Web Science Proceedings of Anagnostopoulos, Aris the 2015 ACM and Becchetti, Luca and SIGMOD The importance of 10.1145/2723372.2 357 Fazzone, Adriano and International 2015 983-998 yes being expert 723722 Mele, Ida and Riondato, Conference on Matteo Management of Data The importance of A. Anagnostopoulos, L. Proceedings of 10.1145/2723372.2 358 being expert: Becchetti, A. Fazzone, I. the 2015 ACM 2015 983-998 yes 723722 efficient max-finding Mele, and M. Riondato SIGMOD in crowdsourcing International Conference on Management of Data Proceedings of the 2015 IEEE Online top-k-position A. Mäcker, M. International monitoring of 10.1109/IPDPS.20 359 Malatyali, and F. M. a. Parallel and 2015 357-364 yes distributed data 15.40 d. Heide Distributed streams Processing Symposium Proceedings of S. Dughmi, A. Eden, M. the 2016 ACM Lottery pricing 10.1145/2940716.2 360 Feldman, A. Fiat, and S. Conference on 2016 401-418 yes equilibria 940742 Leonardi Economics and Computation Proceedings of the 20th ACM Rozenshtein, Polina, SIGKDD Event detection in Aris Anagnostopoulos, international 10.1145/2623330.2 361 2014 yes activity networks Aristides Gionis, and conference on 623674 Nikolaj Tatti Knowledge discovery and data mining Proceedings of Revenue the 24th maximization envy- G. Monaco, P. International 362 free pricing for Sankowski, and Q. 2015 90-96 yes Conference on homogeneous Zhang Artificial resources Intelligence Proceedings of Truthful cake cutting the 24th mechanisms with M. Li, J. Zhang, and Q. International ISBN 978-1- 363 externalities: do not 2015 589-595 yes Zhang Conference on 57735-738-4 make them care for Artificial others too much! Intelligence Anagnostopoulos, Aris and Bessi, Alessandro Proceedings of Viral and Caldarelli, Guido the 24th misinformation: the and Del Vicario Michela 10.1145/2740908.2 364 International 2014 yes role of homophily and Petroni, Fabio and 745939 Conference on and polarization Scala, Antonio and World Wide Web Zollo, Fabiana and Quattrociocchi, Walter Proceedings of Ephemeral networks Akrida, Eleni C and the 26th ACM with random Leszek, G and Mertzios, Symposium on 10.1145/2612669.2 365 availability of links : 2014 yes George B and Spirakis, Parallelism in 612693 diameter and Paul G Algorithms and connectivity Architectures Proceedings of Ephemeral networks E. C. Akrida, L. the 26th ACM with random Gasieniec, G. B. Symposium on 10.1145/2612669.2 366 availability of links: 2014 267-276 yes Mertzios, and P. G. Parallelism in 612693 diameter and Spirakis Algorithms and connectivity Architectures Becchetti, Luca, Andrea Proceedings of Clementi, Emanuele the 26th ACM Simple dynamics for Natale, Francesco Symposium on 10.1145/2612669.2 367 2014 yes plurality consensus Pasquale, Riccardo Parallelism in 612677 Silvestri, and Luca Algorithms and Trevisan Architectures Proceedings of the 26th Annual Plurality consensus ACM-SIAM 368 Becchetti, L 2015 yes in the gossip model Symposium on Discrete Algorithms Proceedings of There is something A. Pacuk, P. Sankowski, the 27th ACM 10.1145/2914586.2 369 beyond the twitter K. Wegrzycki, and P. Conference on 2016 279-284 yes 914623 network Wygocki Hypertext and Social Media 370 Algorithmic P. Brach, M. Cygan, J. Proceedings of 2595 2016 1306- 10.1137/1.9781611 yes complexity of power Lącki, and P. Sankowski the 27th Annual 15 1325 974331.ch91 law networks ACM-SIAM Symposium on Discrete Algorithms Proceedings of Approximately the 27th Annual R. Colini-baldeschi, B. efficient double ACM-SIAM 1424- 10.1137/1.9781611 371 D. Keijzer, S. Leonardi, 2016 yes auctions with strong Symposium on 1443 974331.ch98 and S. Turchetta budget balance Discrete Algorithms Proceedings of the 27th Annual M. Cygan, M. Mucha, Online pricing with ACM-SIAM 10.1137/1.9781611 372 P. Sankowski, and Q. 2016 190-201 yes impatient bidders Symposium on 974331.ch15 Zhang Discrete Algorithms Proceedings of L. Becchetti, A. the 27th Annual Stabilizing consensus Clementi, E. Natale, F. ACM-SIAM 10.1137/1.9781611 373 2016 620-635 yes with many opinions Pasquale, and L. Symposium on 974331.ch46 Trevisan Discrete Algorithms Proceedings of Churn- and dos- the 28th ACM resistant overlay M. Drees, R. Gmyr, and Symposium on 10.1145/2935764.2 374 networks based on 2016 417-427 yes C. Scheideler Parallelism in 935783 network Algorithms and reconfiguration Architectures Proceedings of Facility location with A. Filos-Ratsikas, M. the 29th 375 double-peaked Li, J. Zhang, and Q. Conference on 2015 893-899 arXiv:1502.05548 yes preferences Zhang Artificial Intelligence Proceedings of Spreading rumours the 2nd ACM 10.1145/2660460.2 376 Sankowski, Piotr 2014 yes without the network Conference on 660472 Online Social Networks

Proceedings of the 30th Annual A. Anagnostopoulos, J. Community Conference on Lącki, S. Lattanzi, S. 377 detection on evolving Neural 2016 yes Leonardi, and M. graphs Information Mahdian Processing Systems Proceedings of the 39th International N. A. Alawad, A. Network-aware ACM SIGIR Anagnostopoulos, S. 10.1145/2911451.2 378 recommendations of Conference on 2016 913-916 yes Leonardi, I. Mele, and 914760 novel tweets Research and F. Silvestri Development in Information Retrieval Anagnostopoulos, Aris, Ravi Kumar, Proceedings of Mohammad Mahdian, the 3rd Eli Upfal, Fabio Vandin, Algorithms on Innovations in 10.1145/2090236.2 379 Aris Anagnastopoulos, 2014 149-160 yes evolving graphs Theoretical 090249 Ravi Kumar, Computer Science Mohammad Mahdian, Conference Eli Upfal, and Fabio Vandin Proceedings of the 40th Brach, P, Espasto, A, International Spreading rumors 380 Panconesi, A, Conference on 2013 yes without the network Sankowski, P Automata, Languages, and Programming Strong bounds for Proceedings of Mertzios, George B and 3175 10.1007/978-3- 381 evolution in the 40th 2013 669-680 yes Spirakis, Paul G 32 642-39212-2_58 networks International Conference on Automata, Languages, and Programming Proceedings of Sublinear-time the 40th maintenance of Henzinger, Monika and International breadth-first Krinninger, Sebastian 3175 10.1007/978-3- 382 Conference on 2013 gen-27 yes spanning tree in and Nanongkai, 32 642-39212-2_53 Automata, partially dynamic Danupon Languages, and networks Programming Proceedings of the 40th Temporal network Mertzios, George B and International optimization subject Michail, Othon and 3175 10.1007/978-3- 383 Conference on 2013 657-668 yes to connectivity Chatzigiannakis, Ioannis 32 642-39212-2_57 Automata, constraints and Spirakis, Paul G Languages, and Programming Proceedings of Determining the 41st majority in networks International Mertzios, George B and 10.1007/s00446- 384 with local Colloquium on 6 1081 2014 gen-21 yes Nikoletseas, Sotiris E 016-0277-8 interactions and very Automata small local memory Languages and Programming Proceedings of the 41st Stably computing G. B. Mertzios, S. E. International order statistics with Nikoletseas, C. L. 10.4230/LIPIcs.MF 385 Symposium on 58 2016 gen-68 yes arithmetic population Raptopoulos, and P. G. CS.2016.68 Mathematical protocols Spirakis Foundations of Computer Science Proceedings of Mertzios, George B and Stably computing the 41st Nikoletseas, Sotiris E order statistics with International 10.4230/LIPIcs.MF 386 and Raptopoulos, 69 2016 gen-13 yes arithmetic population Symposium on CS.2016.68 Christoforos L and protocols Mathematical Spirakis, Paul G Foundations of Computer Science Proceedings of the 43rd Reservation G. Goel, S. Leonardi, V. International 142:1- 10.4230/LIPIcs.IC 387 exchange markets for Mirrokni, A. Nikzad, Colloquium on 55 2016 yes 13 ALP.2016.142 internet advertising and R. Paes-Leme Automata, Languages and Programming Space- and time- Proceedings of efficient algorithm S. Bhattacharya, M. the 47th Annual for maintaining Henzinger, D. ACM on 10.1145/2746539.2 388 2015 173-182 yes dense subgraphs on Nanongkai, and C. Symposium on 746592 one-pass dynamic Tsourakakis Theory of streams Computing Proceedings of New deterministic the 48th Annual S. Bhattacharya, M. approximation ACM SIGACT 10.1145/2897518.2 389 Henzinger, and D. 2016 398-411 yes algorithms for fully Symposium on 897568 Nanongkai dynamic matching Theory of Computing Proceedings of Jacobson, Van and the 5th Smetters, Diana K and International Networking named Briggs, Nicholas H and Conference on 10.1145/1658939.1 390 2009 01-dic yes content Thornton, James D and Emerging 658941 Plass, Michael F and Networking Braynard, Rebecca L Experiments and Technologies Temporal multi-layer Proceedings of network construction B. Sluban, M. Grcar, the 7th Workshop 10.1007/978-3- 391 2016 29—41 yes from major news and I. Mozetic on Complex 319-30569-1 events Networks Statistical inference Proceedings of Antulov-Fantulin, Nino framework for source the 8th and Lancic, Alen and detection of International 10.1109/SASOW.2 392 Stefancic, Hrvoje and 2013 yes contagion processes Conference on 014.35 Sikic, Mile and Smuc, on arbitrary network Self-Adaptive and Tomislav structures Self-Organizing Systems Proceedings of Inefficiency of A. Anagnostopoulos, L. the Algorithmic 10.1007/978-3- 393 games with social Becchetti, B. De Game Theory: 6th 2013 219-230 yes 642-41392-6_19 context Keijzer, and G. Schafer International Symposium Bhattacharya, Sayan and Proceedings of Koutsoupias, Elias and Near-optimal multi- the Fourteenth Kulkarni, Janardhan and 10.1145/2482540.2 394 unit auctions with ACM Conference 1 212 2013 91-102 yes Leonardi, Stefano and 482555 ordered bidders on Electronic Roughgarden, Tim and Commerce Xu, Xiaoming Proceedings of the IEEE Monitoring the J. Smailovic, J. Kranjc, International twitter sentiment 10.1109/DSAA.20 395 M. Grcar, M. Znidarsic, Conference on 2015 yes during the Bulgarian 15.7344886 and I. Mozetic Data Science and elections Advanced Analytics Proceedings of the Machine A. Bifet, M. May, B. Learning and Multinets: web-based Zadrozny, R. Gavalda, Knowledge 10.1007/978-3- 396 multilayer network D. Pedreschi, F. Bonchi, 2015 298-302 yes Discovery in 642-40988-2 visualization J. Cardoso, and M. Databases: Spiliopoulou European Conference Elimination of systemic risk in Poledna, Sebastian and Quantitative 1599- 10.1080/14697688. 397 financial networks by 16 10 2014 yes Thurner, Stefan Finance 1613 2016.1156146 means of a systemic risk transaction tax Kenett, Dror Y and Partial correlation Huang, Xuqing and analysis : Quantitative 10.1080/14697688. 398 Vodenska, Irena and 15 4 2015 569-578 yes applications for Finance 2014.946660 Havlin, Shlomo and financial markets Stanley, H Eugene Li, Daqing and Zhang, Network reliability Reliability Qiong and Zio, Enrico 10.1016/j.ress.2015 399 analysis based on Engineering and 142 2015 556-562 yes and Havlin, Shlomo and .05.021 percolation theory System Safety Kang, Rui Pastor-Satorras, Romualdo, Claudio Epidemic processes Reviews of 10.1103/RevModP 400 Castellano, Piet Van 2014 yes in complex networks Modern Physics hys.87.925 Mieghem, and Alessandro Vespignani Approches multiplexes des Revue systèmes de villes d'Economie 10.3917/reru.153.0 401 C. Rozenblat 3 2015 393-424 yes dans les réseaux Régionale et 393 d'entreprises Urbaine multinationales Vodenska, Irena and Becker, Alexander P Community analysis and Zhou, Di and 10.3390/risks40200 402 of global financial Risks 4 2016 13-28 yes Kenett, Dror Y and 13 markets Stanley, H Eugene and Havlin, Shlomo Successful fish go with the flow: Klimek, Peter and S. citation impact Jovanovic Aleksandar 1265- 10.1007/s11192- 403 prediction based on Scientometrics 107 3 2016 yes and Egloff, Rainer and 1282 016-1926-1 centrality measures Schneider, Reto for term-document networks Dynamic approximate all-pairs M. Henzinger, S. shortest paths: SIAM Journal of 947- 10.1137/14095729 404 Krinninger, and D. 45 2016 yes breaking the o(mn) Computing 1006 9 Nanongkai barrier and derandomization Border sensitive Morrison, Greg and Signal-Image centralities in patent Riccaboni, Massimo and Technology & 10.1109/SITIS.201 405 2013 yes citation networks Giovanis, Eleftherios Internet-Based 3.91 using asymmetric and Pammolli, Fabio Systems random walks González-Bailón, Sandra and Wang, Ning Assessing the bias in and Rivero, Alejandro 10.1016/j.socnet.20 406 samples of large Social Networks 38 1 2014 16-27 yes and Borge-Holthoefer, 14.01.004 online networks Javier and Moreno, Yamir Understanding Springer financial news with B. Sluban, J. Smailovic, 10.1007/978-3- 407 International 2016 193-207 yes multi-layer network and I. Mozetic 319-29228-1_17 Publishing analysis Continuous Structural Sebastian Abshoff, aggregation in Information and 10.1007/978-3- 408 Friedhelm Meyer auf 2014 yes dynamic ad-hoc Communication 319-09620-9_16 der Heide networks Complexity Elsa, Bazellières and Vito, Conte and Alberto, Elosgui-Artola and Control of cell-cell Serra-Picamal, Xavier forces and collective and María, Bintanel- 409 cell dynamics by the Morcillo and Roca- Submitted 17 4 2014 10.1038/ncb3135 yes intercellular Cusachs, Pere and José adhesome Muñoz and Marta, Sales-Pardo and Roger, Guimerà and Trepat, Xavier Foschi, Rachele and Missing links in Riccaboni, Massimo and submitted to 10.1109/SITIS.201 410 multiple trade Stefano Schiavo and 2013 yes SITIS 2013 3.96 networks Schiavo, Stefano and Stefano Schiavo A deterministic Kniesburges, Sebastian worst-case message and Koutsopoulos, Theoretical 10.1016/j.tcs.2014. 411 complexity optimal 584 2013 67-79 yes Andreas and Scheideler, Computer Science 11.027 solution for resource Christian discovery On the structure of S. Nikoletseas, P. Theoretical 10.1007/978-3- 412 590 2015 96-105 yes equilibria in basic Panagopoulou, C. Computer Science 642-40164-0_25 network formation Raptopoulos, and P. Spirakis Traveling salesman Michail, Othon and Theoretical 10.1016/j.tcs.2016. 413 problems in temporal 634 2016 gen-23 yes Spirakis, Paul G. Computer Science 04.006 graphs Optimal decremental Łǎcki, Jakub and Theory Computer 10.4230/LIPIcs.ST 414 connectivity in 3 2014 01-ott yes Sankowski, Piotr Systems ACS.2015.608 planar graphs Networks of networks: the last D'Agostino, Gregorio Understanding 10.1007/978-3- 415 2014 yes frontier of and Scala, Antonio Complex Systems 319-03518-5 complexity Anagnostopoulos, Aris and Becchetti, Luca and Markov-chain Castillo, Carlos and optimization for 416 Gionis, Aristides and Cl, Working Paper 2013 gen-28 yes query Chato@chato and recommendation Gionis, Aristides and Castillo, Carlos Uniform price Cigler L., Dvorak W., strategies to exploit 417 Henziger M. Working Paper 2013 yes positive network Starnberger M. externalities A. Magalich, V. Palchykov, V. Gemmetto, D. Sciencewise : topic Garlaschelli, A. modeling over 418 Boyarsky, A. Martini, 2016 yes scientific literature A. Cardillo, A. networks Constantin, O. Ruchayskiy, P. D. L. Rios, and K. Aberer Selection rules in Garas, Antonios, Mario alliance formation: 419 V Tomasello, and Frank 2014 arXiv:1403.3298 yes strategic decisions or Schweitzer abundance of choice?

TEMPLATE A2: LIST OF DISSEMINATION ACTIVITIES

Countries Main Type of addressed NO. Type of activities2 Title Date/Period Place leader audience3 Size of audience 1 A web site for the Proactive initiative 26 February Civil (http://www.dym-cs.eu), IMT Dym-CS 2010 N/A Society N/A International 2 the Conference Session ICALP 2013 in Scientific UPB Riga Latvia June 2013 Riga Community 40 International 3 Horizons in Social Science Scientific Conference IMT June 2013 June 2013 Lucca Community 110 International 4 Ecole de Physique Les Les Scientific School CNRS Houches 2014 April 2014 Houches Community 60 International 5 Scientific Community Civil Conference IMT ECCS’14 September 2014 Lucca Society 850 International 6 Scientific Community Civil Satellite IMT ECCS’14 September 2014 Lucca Society 40 International 7 (G. Networks of D’Agostino, Networks A. Scala Springer Series Scientific Book eds) CNR 2014 N/A Community N/A International 8 Conference UZ NETSCI15 June 2015 Zaragoza Scientific 300 International

2 A drop down list allows choosing the dissemination activity: publications, conferences, workshops, web, press releases, flyers, articles published in the popular press, videos, media briefings, presentations, exhibitions, thesis, interviews, films, TV clips, posters, Other.

3 A drop down list allows choosing the type of public: Scientific Community (higher education, Research), Industry, Civil Society, Policy makers, Medias, Other ('multiple choices' is possible). Community 9 Scientific Session UZ NETSCI15 June 2015 Zaragoza Community 30 International 10 SATELLITE Scientific IMT CCS’15 September 2015 Phoenix Community 30 International 11 Interconnected Scientific Book ETHZ Networks 2015 N/A Community N/A International 12 Data Science and Complex Scientific Book IMT Networks 2015 N/A Community N/A International 13 Scientific Satellite IMT STATPHYS 2016 Paris Community 50 International 14 Scientific WORKSHOP CNR LIPARI 2016 Lipari Community 40 International 15 Scientific SATELLITE IMT/CNR CCS’16 2016 Amsterdam Community 40 International 16 Scientific Community Civil 419 Papers ALL N/A 2012-2016 N/A Society N/A International 17 Multiplex Scientific Book ALL Book 2017 N/A Community N/A International

Section B (Confidential4 or public: confidential information to be marked clearly) Part B1

The applications for patents, trademarks, registered designs, etc. shall be listed according to the template B1 provided hereafter.

The list should, specify at least one unique identifier e.g. European Patent application reference. For patent applications, only if applicable, contributions to standards should be specified. This table is cumulative, which means that it should always show all applications from the beginning until after the end of the project.

TEMPLATE B1: LIST OF APPLICATIONS FOR PATENTS, TRADEMARKS, REGISTERED DESIGNS, ETC.

Confidential Foreseen Click on embargo date YES/NO dd/mm/yyyy Application Type of IP reference(s) Subject or title of Applicant (s) (as on the application) Rights5: (e.g. application EP123456)

4 Note to be confused with the "EU CONFIDENTIAL" classification for some security research projects.

5 A drop down list allows choosing the type of IP rights: Patents, Trademarks, Registered designs, Utility models, Others.

Part B2 Please complete the table hereafter:

Description Confidentia Foreseen Patents or other Type of of exploitable Exploitable Timetable, l embargo Sector(s) of IPR Owner & Other Exploitable foreground product(s) or commercial or Click on date application7 exploitation Beneficiary(s) involved Foreground6 YES/NO dd/mm/yy measure(s) any other use (licences) yy Related FET NO NA NA General project Software tool Policy Not pending ETH Zurich advancement of evaluation, knowledge decision makers support, Trade networks. Multi-Level Policy Evaluation tool (MULTIPOL) Related FET NO NA NA General project Software tool Medicine, Not pending CNR-ISC advancement of cranio-facial knowledge assessment. Orthodontic application of complex network theory (ORTHONET) General Related FET NO NA Software tool Advanced NA Not Pending Institut Jozef stefan advancement of Project knowledge knowledge mining products for investors and traders (FINPLEX)

19 A drop down list allows choosing the type of foreground: General advancement of knowledge, Commercial exploitation of R&D results, Exploitation of R&D results via standards, exploitation of results through EU policies, exploitation of results through (social) innovation. 7 A drop down list allows choosing the type sector (NACE nomenclature) : http://ec.europa.eu/competition/mergers/cases/index/nace_all.html

MULTIPOL – Multi-level Policy Evaluation Tool (Related FET Project) Developed by ETH Zurich, is a software-based tool aimed at evaluating policies that improve the position of a country in the international trade network. The foundations of MULTIPOL lie in the results of MULTIPLEX, whose results on multi-layer networks are transferred into a system supporting decisions on country specific portfolios of commodities produced. Weighted and directed links represent trade relations, while different layers of the network represent commodities. MULTIPOL defines a set of measures defining network resilience and the vulnerability of the network, giving insights about: 1) The impact of decisions to increase/decrease production of certain commodities at country level; 2) The impact of decision to start producing new ones; 3) The network position of a country and the global resilience of the international trade network.

ORTHONET - Orthodontic application of complex network theory (Related FET project) Developed by CNR-ISC, is a software application devoted to medical doctors working in the field of orthodontic and cranio-facial field. Cephalometric variables can be adopted to assess the cranio-facial situation of different patients. Starting from longitudinal data from single patients and combining with cross-sectional ones (data averaged on a population of patients), ORTHONET is able to provide doctors with a plausible evolution of patients’ malocclusion. Such hypothesis is tested by use of Bayesian staistics in order to quantify the probability of success for medical treatments.

FINPLEX - Advanced Knowledge Mining Products for Investors and Traders (Related FET Project) Presented by Institut Jozef Stefan. In the last few years, investor sentiment (“stance”) in social media has become an important indicator for financial decision-making complementing news, fundamental, and technical analyses. The aim of FINPLEX is to apply sentiment analysis to the financial domain, as large financial data providers have already entered the market. The project exploits the advanced algorithms, techniques, and platforms produced by MULTIPLEX, and we propose to take these results as the foundation of new products and services to be offered within Europe. A further goal of FINPLEX is to verify and substantiate the potential of ideas arising from MULTIPLEX, and to take necessary steps towards successful commercialisation, including establishing customer channels and running pilot cases, and preparing a thorough business plan that will be the basis for seeking investment funding after the end of the project. In the long run, it will enable researchers to enter the business world. be. 4.3 Report on societal implications

Replies to the following questions will assist the Commission to obtain statistics and indicators on societal and socio-economic issues addressed by projects. The questions are arranged in a number of key themes. As well as producing certain statistics, the replies will also help identify those projects that have shown a real engagement with wider societal issues, and thereby identify interesting approaches to these issues and best practices. The replies for individual projects will not be made public.

A General Information (completed automatically when Grant Agreement number is entered.

Grant Agreement Number: 317532

Title of Project: MULTIPLEX

Name and Title of Coordinator: PROF. GUIDO CALDARELLI B Ethics

1. Did your project undergo an Ethics Review (and/or Screening)?

• If Yes: have you described the progress of compliance with the relevant Ethics NO Review/Screening Requirements in the frame of the periodic/final project reports?

Special Reminder: the progress of compliance with the Ethics Review/Screening Requirements should be described in the Period/Final Project Reports under the Section 3.2.2 'Work Progress and Achievements'

2. Please indicate whether your project involved any of the following issues (tick NO box) : RESEARCH ON HUMANS • Did the project involve children? • Did the project involve patients? • Did the project involve persons not able to give consent? • Did the project involve adult healthy volunteers? • Did the project involve Human genetic material? • Did the project involve Human biological samples? • Did the project involve Human data collection? RESEARCH ON HUMAN EMBRYO/FOETUS • Did the project involve Human Embryos? • Did the project involve Human Foetal Tissue / Cells? • Did the project involve Human Embryonic Stem Cells (hESCs)? • Did the project on human Embryonic Stem Cells involve cells in culture? • Did the project on human Embryonic Stem Cells involve the derivation of cells from Embryos? PRIVACY • Did the project involve processing of genetic information or personal data (eg. health, sexual lifestyle, ethnicity, political opinion, religious or philosophical conviction)? • Did the project involve tracking the location or observation of people? RESEARCH ON ANIMALS • Did the project involve research on animals? • Were those animals transgenic small laboratory animals? • Were those animals transgenic farm animals? • Were those animals cloned farm animals? • Were those animals non-human primates? RESEARCH INVOLVING DEVELOPING COUNTRIES • Did the project involve the use of local resources (genetic, animal, plant etc)? • Was the project of benefit to local community (capacity building, access to healthcare, education etc)? DUAL USE • Research having direct military use NO • Research having the potential for terrorist abuse C Workforce Statistics 3. Workforce statistics for the project: Please indicate in the table below the number of people who worked on the project (on a headcount basis). Type of Position Number of Women Number of Men

Scientific Coordinator 0 1 Work package leaders 0 4 Experienced researchers (i.e. PhD holders) 26 10 PhD Students 0 65 Other 4. How many additional researchers (in companies and universities) were 67 recruited specifically for this project? Of which, indicate the number of men: 45

D Gender Aspects 5. Did you carry out specific Gender Equality Actions under the project? X Yes ¡ No 6. Which of the following actions did you carry out and how effective were they? Not at all Very effective effective q Design and implement an equal opportunity policy ¡ ¡ ¡ ¡ ¡ q Set targets to achieve a gender balance in the workforce ¡ ¡ ¡ ¡ ¡ q Organise conferences and workshops on gender ¡ ¡ ¡ ¡ ¡ q Actions to improve work-life balance ¡ ¡ X ¡ ¡ ¡ Other: IN CASE OF SIMILAR CANDIDATES CONSIDERED GENDER ISSUE

7. Was there a gender dimension associated with the research content – i.e. wherever people were the focus of the research as, for example, consumers, users, patients or in trials, was the issue of gender considered and addressed? ¡ Yes- please specify

X No E Synergies with Science Education

8. Did your project involve working with students and/or school pupils (e.g. open days, participation in science festivals and events, prizes/competitions or joint projects)? ¡ Yes- please specify

X No 9. Did the project generate any science education material (e.g. kits, websites, explanatory booklets, DVDs)? ¡ Yes- please specify

X No F Interdisciplinarity

10. Which disciplines (see list below) are involved in your project? ¡ Main discipline8: X Associated discipline8: 1.1; X Associated discipline8: 1.2; 5.2;

G Engaging with Civil society and policy makers 11a Did your project engage with societal actors beyond the research X Yes community? (if 'No', go to Question 14) ¡ No 11b If yes, did you engage with citizens (citizens' panels / juries) or organised civil society (NGOs, patients' groups etc.)? X No ¡ Yes- in determining what research should be performed ¡ Yes - in implementing the research ¡ Yes, in communicating /disseminating / using the results of the project

8 Insert number from list below (Frascati Manual). ¡ Yes 11c In doing so, did your project involve actors whose role is mainly to X No organise the dialogue with citizens and organised civil society (e.g. professional mediator; communication company, science museums)? 12. Did you engage with government / public bodies or policy makers (including international organisations)

¡ No X Yes- in framing the research agenda “Reducing systemic risk” ¡ Yes - in implementing the research agenda ¡ Yes, in communicating /disseminating / using the results of the project 13a Will the project generate outputs (expertise or scientific advice) which could be used by policy makers? X Yes – as a primary objective (please indicate areas below- multiple answers possible) ¡ Yes – as a secondary objective (please indicate areas below - multiple answer possible) ¡ No 13b If Yes, in which fields? ENERGY, ECONOMICS, FINANCE Agriculture Energy Human rights Audiovisual and Media Enlargement Information Society Budget Enterprise Institutional affairs Competition Environment Internal Market Consumers External Relations Justice, freedom and security Culture External Trade Public Health Customs Fisheries and Maritime Affairs Regional Policy Development Economic and Food Safety Research and Innovation Monetary Affairs Foreign and Security Policy Space Education, Training, Youth Fraud Taxation Employment and Social Affairs Humanitarian aid Transport

13c If Yes, at which level? ¡ Local / regional levels ¡ National level X European level ¡ International level H Use and dissemination

14. How many Articles were published/accepted for publication in 419 peer-reviewed journals? To how many of these is open access9 provided? ALL How many of these are published in open access journals? 257 How many of these are published in open repositories? ALL To how many of these is open access not provided? 0

Please check all applicable reasons for not providing open access: q publisher's licensing agreement would not permit publishing in a repository

q no suitable repository available q no suitable open access journal available q no funds available to publish in an open access journal q lack of time and resources q lack of information on open access q other10: …………… 15. How many new patent applications (‘priority filings’) have been made? 0 ("Technologically unique": multiple applications for the same invention in different jurisdictions should be counted as just one application of grant). 16. Indicate how many of the following Intellectual Trademark 0 Property Rights were applied for (give number in each box). Registered design 0 Other 0

17. How many spin-off companies were created / are planned as a direct 2 result of the project? Indicate the approximate number of additional jobs in these companies: 5 18. Please indicate whether your project has a potential impact on employment, in comparison with the situation before your project: q Increase in employment, or q In small & medium-sized enterprises q Safeguard employment, or q In large companies q Decrease in employment, q None of the above / not relevant to the project X Difficult to estimate / not possible to quantify 19. For your project partnership please estimate the employment effect Indicate figure: resulting directly from your participation in Full Time Equivalent (FTE =

one person working fulltime for a year) jobs:

9 Open Access is defined as free of charge access for anyone via Internet. 10 For instance: classification for security project.

Difficult to estimate / not possible to quantify X

I Media and Communication to the general public

20. As part of the project, were any of the beneficiaries professionals in communication or media relations? X Yes ¡ No 21. As part of the project, have any beneficiaries received professional media / communication training / advice to improve communication with the general public? ¡ Yes X No 22 Which of the following have been used to communicate information about your project to the general public, or have resulted from your project? X Press Release X Coverage in specialist press q Media briefing q Coverage in general (non-specialist) press q TV coverage / report q Coverage in national press q Radio coverage / report q Coverage in international press q Brochures /posters / flyers q Website for the general public / internet q DVD /Film /Multimedia q Event targeting general public (festival, conference, exhibition, science café) 23 In which languages are the information products for the general public produced? q Language of the coordinator X English q Other language(s)

Question F-10: Classification of Scientific Disciplines according to the Frascati Manual 2002 (Proposed Standard Practice for Surveys on Research and Experimental Development, OECD 2002):

FIELDS OF SCIENCE AND TECHNOLOGY

1. NATURAL SCIENCES 1.1 Mathematics and computer sciences [mathematics and other allied fields: computer sciences and other allied subjects (software development only; hardware development should be classified in the engineering fields)] 1.2 Physical sciences (astronomy and space sciences, physics and other allied subjects) 1.3 Chemical sciences (chemistry, other allied subjects) 1.4 Earth and related environmental sciences (geology, geophysics, mineralogy, physical geography and other geosciences, meteorology and other atmospheric sciences including climatic research, oceanography, vulcanology, palaeoecology, other allied sciences) 1.5 Biological sciences (biology, botany, bacteriology, microbiology, zoology, entomology, genetics, biochemistry, biophysics, other allied sciences, excluding clinical and veterinary sciences)

2 ENGINEERING AND TECHNOLOGY 2.1 Civil engineering (architecture engineering, building science and engineering, construction engineering, municipal and structural engineering and other allied subjects) 2.2 Electrical engineering, electronics [electrical engineering, electronics, communication engineering and systems, computer engineering (hardware only) and other allied subjects] 2.3. Other engineering sciences (such as chemical, aeronautical and space, mechanical, metallurgical and materials engineering, and their specialised subdivisions; forest products; applied sciences such as geodesy, industrial chemistry, etc.; the science and technology of food production; specialised technologies of interdisciplinary fields, e.g. systems analysis, metallurgy, mining, textile technology and other applied subjects)

3. MEDICAL SCIENCES 3.1 Basic medicine (anatomy, cytology, physiology, genetics, pharmacy, pharmacology, toxicology, immunology and immunohaematology, clinical chemistry, clinical microbiology, pathology) 3.2 Clinical medicine (anaesthesiology, paediatrics, obstetrics and gynaecology, internal medicine, surgery, dentistry, neurology, psychiatry, radiology, therapeutics, otorhinolaryngology, ophthalmology) 3.3 Health sciences (public health services, social medicine, hygiene, nursing, epidemiology)

4. AGRICULTURAL SCIENCES 4.1 Agriculture, forestry, fisheries and allied sciences (agronomy, animal husbandry, fisheries, forestry, horticulture, other allied subjects) 4.2 Veterinary medicine

5. SOCIAL SCIENCES 5.1 Psychology 5.2 Economics 5.3 Educational sciences (education and training and other allied subjects) 5.4 Other social sciences [anthropology (social and cultural) and ethnology, demography, geography (human, economic and social), town and country planning, management, law, linguistics, political sciences, sociology, organisation and methods, miscellaneous social sciences and interdisciplinary , methodological and historical S1T activities relating to subjects in this group. Physical anthropology, physical geography and psychophysiology should normally be classified with the natural sciences].

6. HUMANITIES 6.1 History (history, prehistory and history, together with auxiliary historical disciplines such as archaeology, numismatics, palaeography, genealogy, etc.) 6.2 Languages and literature (ancient and modern) 6.3 Other humanities [philosophy (including the history of science and technology) arts, history of art, art criticism, painting, sculpture, musicology, dramatic art excluding artistic "research" of any kind, religion, theology, other fields and subjects pertaining to the humanities, methodological, historical and other S1T activities relating to the subjects in this group]

2. FINAL REPORT ON THE DISTRIBUTION OF THE EUROPEAN UNION FINANCIAL CONTRIBUTION

Report on the distribution of the European Union financial contribution between beneficiaries