review articles

doi:10.1145/2699416 Jacob Moreno’s25 efforts to develop Exploring three interdisciplinary areas and the “sociometry.” Soon thereafter, the mathematical framework offered by extent they overlap. Are they all part of the graph theory was also picked up by same larger domain? psychologists,2 anthropologists,23 and other social scientists to create an in- By Thanassis Tiropanis, , Jon Crowcroft, terdiscipline called Social Networks. Noshir Contractor, and Leandros Tassiulas The interdiscipline of Social Networks expanded even further toward the end of the 20th century with an explosion of interest in exploring networks in biological, physical, and technologi- Network cal systems. The term Network Sci- ence emerged as an interdisciplinary area that draws on disciplines such as physics, mathematics, computer sci- Science, ence, biology, economics, and sociol- ogy to encompass networks that were not necessarily social.1,26,35 The study of networks involves developing ex- planatory models to understand the , emergence of networks, building predictive models to anticipate the evolution of networks, and construct- ing prescriptive models to optimize and the outcomes of networks. One of the main tenets of Network Science is to identify common underpinning principles and laws that apply across Science very different networks and explore why in some cases those patterns vary. The Internet and the Web, given their spectacular growth and impact, are networks that have captured the imagination of many network scien-

The observation of patterns that characterize key insights networks, from biological to technological and social, ˽˽ Web Science and Internet Science aim to understand the evolution of the Web and the impact of the Web and the Internet on society and the Internet respectively and to inform debates about their future. These and business have motivated interdisciplinary research goals lead to different priorities in their to advance our understanding of these systems. Their research agendas even though their communities over overlap.

study has been the subject of Network Science research ˽˽ Network Science aims to understand the evolution of networks regardless for a number of years. However, more recently we have of where they emerge: the Internet as a witnessed the emergence of two new interdisciplinary network transforming and forwarding information among people and things, areas: Web Science and Internet Science. and the Web as a network of creation and Network Science can be traced to its mathematical . ˽˽ Given their intellectual complementarities, origins dating back to Leonard Euler’s seminal work we propose sharing and harmonizing 15 th the data research infrastructures on graph theory in the 18 century and to its social being developed across these three interdisiplinary communities. scientific origins two centuries later by the psychiatrist credit tk

1 communications of the acm | august 2015 | vol. 58 | no. 8 august 2015 | vol. 58 | no. 8 | communications of the acm 2 review articles

tists.13 In addition, the emergence of and sociology as well as computer sci- security, privacy, and resilience. This online social networks and the poten- ence and engineering. A major focus motivates the need for multidisci- tial to study online interactions on a of the Web Science research agenda is plinary research on Internet Science massive, global scale hold the prom- to understand how the Web is evolv- that seeks to understand the psycho- ise of further, potentially invaluable ing as a socio-technical phenomenon logical, sociological, and economic insights to network scientists on net- and how we can ensure it will contin- implications of the Internet’s evolu- work evolution.24 ue to evolve and benefit society in the tion along these principled direc- Web Science6 is an interdisciplinary years to come. tions. Hence, Internet Science is an area of much more recent vintage that Internet Science. The Internet has emerging interdisciplinary area that studies the Web not only at the level of provided the infrastructure on which brings together scientists in network small technological innovations (mi- much of human activity has become engineering, computation, complexi- cro level) but also as a phenomenon heavily dependent. After only a few ty, security, trust, mathematics, phys- that affects societal and commercial decades of Internet development it ics, sociology, economics, political activities globally (macro level); to a is self-evident that if the Internet be- sciences, and law. This approach is large extent, it can be considered the came unavailable, the consequences very well exemplified by the early In- theory and practice of social machines for society, commerce, the economy, ternet topology study.16 on the Web. Social machines were defense, and government would be Interdisciplinary relationships. All conceptualized by Tim Berners-Lee in highly disruptive. The success of the three areas draw on a number of dis- 1999 as artifacts where people do the Internet has often been attributed ciplines for the study, respectively, creative work and machines interme- to its distributed governance model, the nature and impact of the Web, diate.3 and the principle of network neutrality, the Internet and networks in general technologies can provide the means and its openness.14 At the same time, on government, business, people, for knowledge representation and rea- concerns related to privacy, securi- devices, and the environment. How- soning and enable further support for ty, openness, and sustainability are ever, each of them examines how social machines.20 raised and researched as they are of- those actors co-create and evolve in Studying the Web and its impact re- ten at the center of contestations on distinct, unique ways as shown on quires an interdisciplinary approach the Internet.11 The Internet can be Figure 1. For Web Science it is the that focuses not only on the techno- seen as an infrastructure, the social aspect of linking those actors and logical level but also on the societal, value of which must be safeguard- the content with which they interact political and commercial levels. Es- ed.18 It is the infrastructure that en- making associations between them tablishing the relationship between abled the evolution of the Web along and interpreting them. For Internet these levels, understanding how they with P2P applications, more recently Science it is the aspect of communi- influence each other, investigating the cloud, and, in the near future, the cation among actors and resources potential underpinning laws and ex- Internet of Things. It has been ar- as processes that can shape informa- ploring ways to leverage this relation- gued the infrastructural layer of the tion relay and transformation. For ship in different domains of human Internet and that of the Web must be Network Science, it is the aspect of activity is a large part of the Web Sci- kept separately to foster innovation.4 how these entities, when considered ence research agenda. Web Science A recent study7 identified a number to be part of a network, exhibit cer- draws on disciplines that include the of principled directions along which tain characteristics and might ad- social sciences, such as anthropology, the Internet needs to evolve; those here to underpinning laws that can communication, economics, law, phi- include availability, inclusiveness, help understand their evolution. losophy, political science, psychology, scalability, sustainability, openness, However, to understand better the

Figure 1. Web, Internet, and Network Science aspects.

Social Sciences Web Science People Engineering The content (co)creation, linkage and evolution aspect Web protocols, code and policies

Computer Science Government

Psychology Internet Science Business Law The information relay and transformation aspect Internet protocols, code and policies

Economics Devices

Education Network Science Environment Network properties and network evolution Mathematics

…

3 communications of the acm | august 2015 | vol. 58 | no. 8 review articles similarities and differences between cial scientists were joined by a large these areas and to establish the po- and growing influx of scholars from tential for synergies, a framework for the physical and life sciences who be- a more detailed comparison is needed. gan exploring networks in social sys- tems. This effort was acknowledged A Comparison of All three areas and further catalyzed by the launch Interdisciplinary Areas draw on a number of annual Network Science (NetSci) It takes only a quick read through a conference in 2006, a major infusion short description of each of these in- of disciplines for of funding in 2008 from the Army terdisciplinary areas5,32,35 for one to the study, Research Laboratory for the develop- realize that, to a very large extent, they ment of an interdisciplinary Network all draw from very similar sets of dis- respectively, Science Collaborative Technology Al- ciplines. Venn diagrams that have liance (NS-CTA), and the launch of been used to illustrate the involve- the nature and Network Science in 2013. Clearly there ment of different disciplines in each impact of the Web, was already a community in place, area are indicative of this overlap. For which engaged in interdisciplinary example, psychology and economics the Internet and work long before those initiatives; are considered relevant to Network networks in general one can argue a hybrid bottom-up and Science,29 Internet Science,7 and Web top-down approach is the community Science.20 This can give rise to certain on government, formation model that was followed questions such as: “If there is so much business, people, for Network Science. overlap, aren’t these areas one and the For the Web Science community, same?” or “Would they all merge in devices, and it was around 2006 when it was real- the future?” Other questions include: the environment. ized that understanding the impact of “Which community is more relevant to the Web was essential to safeguard its my research?” or “What developments development in the future. The Web could we expect from each area in the Science Research Initiative (WSRI) future?” To explore those questions was established in 2006 and later de- we propose a framework of examining veloped into the Web Science Trust those interdisciplinary areas, which (WST) as part of the top-down ap- includes looking at the way that these proach to the formation of the Web communities have formed, and the Science community. The WSRI raised different languages of discourse that a banner for those who were engaged these communities have employed in in research on the Web as a socio- their research. technical phenomenon, including Community formation. Although the social network research commu- not all three interdisciplinary do- nity. A similar community formation mains were established at the same model was followed for Internet Sci- time, one can argue that research in ence, where the European Network of those areas dates back before their of- Excellence in Network Science (EINS)a ficial starting date. At the same time, is one of the most significant activi- one can also argue there are differenc- ties to bring together the research es in how communities around those community in this area. Areas such as domains emerged. privacy and network neutrality have The formation of the Social Net- been highlighted as priorities in the works community can be traced back Internet Science agenda. to a series of social network confer- It can be argued the top-down ences that started in the 1970s17 with model of community formation can an important conference in Dart- accelerate research in emergent in- mouth in 1975 that brought together terdisciplinary areas but, in order to sociologists, anthropologists, social be successful, it requires a signifi- psychologists, and mathematicians cant investment of resources from from the U.S. and Europe. This was individuals, from research institu- followed by Lin Freeman’s launch tions, and from industry or govern- of Social Networks in 1978, and Barry ment. Although the Web Science Wellman founding the International and the Internet Science commu- Network for Social Network Analysis nities were formed mostly in a top- (INSNA) in 1976 and its annual Sun- down fashion, the sustainability of belt Social Networks conference in 1981. Beginning in the 1990s, the so- a http://www.internet-science.eu

august 2015 | vol. 58 | no. 8 | communications of the acm 4 review articles

those communities was ensured by understanding of the principles of future and the research methods or research funding from key research Internet protocols, infrastructure data will continue to mingle between institutions, national research coun- (routers, links, AS topology), social these three areas. For example, data cils, the European Union, and signifi- science (preferential attachment on the Internet of Things might not cant effort by individuals. models), law, and policy. remain exclusive to Internet Science Use of a lingua franca. Beyond Research methodologies. In Net- since that data could be combined community formation, there are dif- work Science, research methodolo- with data on human behavior on the ferences in the language of discourse gies involve network modeling and Web from the Web Science perspec- (the lingua franca) that is employed in network analysis9,10 on networks that tive or to explore emergence and out- each area. Network scientists initially include, but are by no means restrict- comes of the networks they enable shared graph theory as their lingua ed to, the Web and the Internet. In from the Network Science point of franca but have more recently em- Internet Science, methodologies that view. Similarly, data on the behavior ployed models taken from physical employ measurements of engage- of users on the Web will be used to processes (percolation, diffusion) and ment of Internet users with online explore the use of bandwidth in the game theory13 to describe processes resources and the Internet of Things underlying Internet infrastructure. on graphs. They have also moved are prevalent. In Web Science, mixed The different types of measurement from descriptive network metrics to research methods that combine inter- point to the fact that often, part of the the development of novel inferential pretative and positivist approaches research, especially in the top-down- techniques to test hypotheses about are employed widely to understand formed areas of Internet Science and the evolution of a network based the evolution of the Web based on Web Science, are associated to spe- on various self-organizing mecha- online social network datasets, click- cific goals. nisms.27,28 As a result, the use of graph stream behaviour, and the use of the Given this shared pool of methods theory is not necessarily the founda- Web of data. and data resources, each area em- tion for contemporary Network Sci- Beyond methodologies, the Web ploys mixed methods to leverage this ence research. Further, there is use of Science community is working on pool in different ways according to complex systems analysis to deal with providing the Web Science Observa- their research agendas as illustrated phase changes and discontinuities tory,33,34 a global distributed resource in Figure 2. Those agendas are in- between different operating regimes; with datasets and analytic tools re- formed by different research goals. these are used to study why epidemic lated to Web Science. Similarly, the Research goals. “Web Science is and pandemics spread globally. As a EINS project is working on provid- focused on how we could do things result, many Network Science publi- ing an evidence base for Internet better, while Network Science is more cations are featured in journals such Science research. And the Network focused on how things work;”36 the as Nature. Science community have a long tra- “doing things better” refers to lever- The Web Science community has dition of making canonical network aging the potential of the Web and not yet embraced a lingua franca datasets available for use by the com- ensuring its continuing sustainabil- per se but one can argue that an un- munity along with network analysis ity. Similar claims are made on be- derstanding of Web standards, tech- software such as UCINET8 and large- half of Internet Science and Network nologies, and models (HTTP, XML, scale repositories of network data Science. Although the use of the term JavaScript, REST, models of commu- such as SNAP.22 ‘science’ relates to the systematic or- nication, ontologies) and of frame- Clearly there is an overlap in the re- ganization of knowledge and is not works of social theory are compo- search methodologies of these three directly linked to goals, we argue that nents of what could develop into a areas: goals do play a role in the formation lingua franca. The W3C has been fos- ˲˲ They draw on data gathered from of these interdisciplinary areas and tering a significant part of the discus- social networks, infrastructures, sen- in shaping their research agendas, sion on Web protocols and their im- sors and the Internet of Things, scientific contribution, and impact. plications. A basic understanding of ˲˲ They involve measurement, mod- In Web Science, the study of the Web the evolution of the Web on both the elling, simulation, visualisation, hy- itself is crucial,21 as is safeguard- micro and macro levels is the founda- pothesis testing, interpretation and ing the Web and its evolution.19 In tion for Web Science research. exploratory research, and Internet Science, the evolution and Similar means of discourse are ˲˲ They use analytical techniques to sustainability of the Internet and its employed in the Internet Science quantify properties of a network (ab- services are central objectives; it is community. For Internet Science, stract, virtual or real) as well as more understood that tussles will always the components of the lingua franca qualitative techniques. be the case on the Internet and that include the set of Internet standards So far, there has been significant accommodating them is necessary (RFCs) and associated commentary emphasis on the social sciences in in order to ensure its evolution.11 It and implementation (or even C code) Web Science, on both social science seems that in both Internet Science as in Stevens’ books,30,31 as well as theories and methodologies in Net- and Web Science applied research the existence of de facto standard work Science and on protocols and comes first but it should be informed implementations of systems in open computer science in Internet Science. by the development of a basic re- source. They also include a basic However, these foci will change in the search program. In addition, neither

5 communications of the acm | august 2015 | vol. 58 | no. 8 review articles

Web Science nor Internet Science is sarily relate to Internet or Web science. work science techniques. technology neutral; each one relies 4. Web Science and Internet Science: 5. Internet Science and Network Sci- on specific protocols and standards. Network neutrality is an example that ence: Content Delivery Networks can Further, one can argue that even the requires understanding of both Web require network techniques for distri- code that implements those stan- and Internet technology and it does bution prediction and optimization dards embeds policy on which each not necessarily draw primarily on net- and, at the same time, understanding respective community has reached some consensus. On the other hand, Figure 2. Network, Internet, and Web Science methodologies. Network Science is technology agnos- tic and it overlaps only in part with Internet Science and Web Science, Internet Science: Web Science: Network Science: Internet Engineering Web Infrastructure Underpinning network laws since it explores emergent structural Internet Evolution Social machines Scale-free network evolution patterns and flows on network struc- Web Evolution tures be they social, biological, the Web, or the Internet. Finally, Web Science and Internet Science are QUALITATIVE QUANTITATIVE both also engineering disciplines; they are about building better, stron- MEASUREMENT, MODELLING, ger, more robust, efficient, and re- SIMULATION, VISUALIZATION, silient systems. Network Science has INTERPRETATION, HYPOTHESIS TESTING been predominantly focused on un- derstanding and describing emergent processes although, access to large DATA media datasets have increased interest in social networks both predictive analytics to antici- network usage device data pate network changes and prescrip- open data tive analytics to optimize networks to archives accomplish certain desired goals. In environment data essence, Network Science is aspiring to take insights from basic research to engineer better networks.12 Figure 3. Research topics differentiating areas and overlaps. Comparisons. Despite the differ- ences between these areas in terms of the community formation mod- els, lingua francas, and goals, many of the research methods they employ Network are common. This points to potential Science synergies on topics in which these ar- 3 eas overlap and the potential for mo- e.g. transport networks bilization within those communities on topics in which there is little or no overlap. Figure 3 shows such topics 6 from each of these areas: 5 1. Web Science: The area of Web- e.g. CDN e.g. diffusion based is one example of on Twitter primarily Web Science research. Net- work aspects are not the exclusive part e.g. SOPA of this since social media research trust focuses on associations and interac- 7 tion among people and social media resources. e.g. net 2. Internet Science: Research on how neutrality the Internet of Things affects informa- e.g. loT 4 e.g. social media tion collection and transformation is 2 primarily Internet Science research 1 Internet Web that cannot rely exclusively on network Science Science research either. 3. Network Science: Transport net- works provide an example of network science research that does not neces-

august 2015 | vol. 58 | no. 8 | communications of the acm 6 review articles

of how Internet protocols and people laboration to harmonize and share re- 22. Leskovec, J. Stanford large network dataset collection, 2011; http://snap.stanford.edu/data/index.html relate to shaping that demand. sources; this should be a high priority 23. Mitchell, J.C. The Kalela Dance; Aspects of Social 6. Network Science and Web Science: for researchers and funding agencies. Relationships among Urban Africans. Manchester University Press, 1956. Diffusion on social media such as Twit- Acknowledgments. The prepara- 24. Monge, P.R. and Contractor, N.S. Theories of ter is an example that relies on Web tion of this manuscript was supported Communication Networks. Oxford University Press, 2003. 25. Moreno, J.L. Who Shall Survive? Foundations of Science socio-technical research meth- by funding from the U.S. Army Re- Sociometry, Group Psychotherapy and Socio-Drama ods and, at the same time, on network search Laboratory (9500010212/0013// (2nd ed.). Beacon House, Oxford, England, 1953. 26. Newman, M.E.J. Networks: An Introduction. Oxford analytic methods. W911NF-09-2-0053), the National Sci- University Press, 2010. 7. Web Science, Internet Science and ence Foundation (CNS-1010904) and 27. robins, G., Snijders, T., Wang, P., Handcock, M., and Pattison, P. Recent developments in exponential Network Science: Research on trust on- the European Union FP7 Network of random graph (p*) models for social networks. Social Networks 29, 2 (2007), 192–215; doi:10.1016/j. line or on SOPA (Stop Online Piracy Excellence in Internet Science–EINS socnet.2006.08.003 Act) and its side effects draw on all (grant agreement No 288021). The 28. Snijders, T.A.B., Van de Bunt, G.G., and Steglich, C.E.G. Introduction to stochastic actor-based models for networks and on techniques that are views, opinions, and/or findings con- network dynamics. Social Networks 32, 1 (2010). aware of Web, and Internet protocols tained here are those of the authors, 44–60; doi:10.1016/j.socnet.2009.02.004 29. Steen, M.V. Computer science, informatics, and the and code. and should not be construed as an of- networked world. Internet Computing, IEEE 15, 3 As the Web and the Internet con- ficial Department of the Army, NSF or (2011), 4–6. 30. Stevens, W.R. TCP/IP illustrated, Vol. 1: The protocols, tinue to evolve it could be that some of European Commission position, poli- (1993). these topics will shift. cy, or decision, unless so designated by 31. Stevens, W.R. and Wright, G.R. TCP/IP Illustrated, Vol. 2: The Implementation, (1995). other documents. 32. the EINS Consortium. EINS Factsheet. The EINS Conclusion Network of Excellence, EU-FP7 (2012); http://www. internet-science.eu/publication/231. We provide a comparison among Net- References 33. tiropanis, T., Hall, W., Hendler, J. de Larrinaga, work, Web, and Internet Science. We 1. Barabási, A.-L. and Albert, R. Emergence of scaling in Christian. The Web Observatory: A middle layer random networks. Science 286, 5439 (1999), 509–512. for broad data. Dx.Doi.org 2, 3 (2014), 129–133; also propose a framework for compar- 2. Bavelas, A. Communication patterns in task-oriented doi:10.1089/big.2014.0035 ing interdisciplinary areas based on groups. J. Acoustical Society of America 22, 6 (1950), 34. tiropanis, T., Hall, W., Shadbolt, N., de Roure, D., 725–730. Contractor, N. and Hendler, J. The Web Science their community formation, lingua 3. Berners-Lee, T. Weaving the Web. Texere Publishing, Observatory. Intelligent Systems, IEEE 28, 2 (2013), francas, research methods and re- 1999. 100–104. 4. Berners-Lee, T. Long live the Web. Scientific American, 35. Watts, D. The ‘new’ science of networks. Annual sources, and research goals. We can (2010). Review of Sociology, (2004). gain additional insights of the rela- 5. Berners-Lee, T., Hall, W., Hendler, J.A., O’Hara, K. 36. Wright, A. Web science meets network science. and Shadbolt, N. A framework for Web science. Commun. ACM 54, 5 (May 2011). tionship among these areas by con- Foundations and Trends in Web Science 1. 1 (2006b), ducting co-author and co-citation 1–130. Thanassis Tiropanis ([email protected]) is 6. Berners-Lee, T., Hall, W., Hendler, J., Shadbolt, N. and an associate professor with the Web and Internet Science analysis of publications within these Weitzner, D. Computer science enhanced: Creating a Group, Electronics and Computer Science, University of areas and explore the extent to which science of the Web. Science 313, 5788 (2006a), 769. Southampton, U.K. 7. Blackman, C., Brown, I., Cave, J., Forge, S., Guevara, these are distinct or merging inter- K., Srivastava, L. and Popper, M.T.W.R. Towards a Wendy Hall ([email protected]) is a professor of Computer Science at the University of Southampton, U.K. disciplinary intellectual communi- Future Internet, (2010). European Commission DG INFSO Project SMART 2008/0049. She is a former president of ACM. ties. Such an analysis would be even 8. Borgatti, S., Everett, M.G. and Freeman, L.C. Computer Jon Crowcroft ([email protected]) is the more meaningful as the related con- Software: UCINET 6. Analytic Technologies, 2006. 9. Börner, K., Sanyal, S. and Vespignani, A. Network Marconi Professor of Communications Systems in the ferences and journals mature and as science B. Cronin (ed.). Annual Review of Information Computer Lab, at the University of Cambridge, U.K. Science & Technology, 41 (2007), 537–607. the similarities and differences among 10. carrington, P. J., Scott, J., and Wasserman, S., eds. Noshir Contractor ([email protected]) is the Jane these areas potentially crystallize. Models and Methods in Social Network Analysis. S. & William J. White Professor of Behavioral Sciences in Cambridge University Press, 2005. the McCormick School of Engineering & Applied Science, Both Internet Science and Web Sci- 11. clark, D.D., Wroclawski, J., Sollins, K.R., and Braden, the School of Communication, and the Kellogg School of ence are technology-aware and their R. Tussle in cyberspace: Defining tomorrow’s Internet. Management at Northwestern University, Chicago, IL. Aug. 2002. ACM. respective lingua francas include 12. contractor, N.S. and DeChurch, L.A. (in press). Leandros Tassiulas ([email protected]) is the knowledge of the protocols and sys- Integrating social networks and human social motives John C. Malone Professor of Electrical Engineering at Yale to achieve social influence at scale. In Proceedings of University, New Haven, CT. tems supporting the Internet and the the National Academy of Sciences. Web, while Network Science is technol- 13. easley, D. and Kleinberg, J. Networks Crowds and © 2015 ACM 0001-0782/15/07 $15.00 Markets. Cambridge University Press, 2010. ogy-agnostic. There are arguments in 14. economides, N. Net neutrality, non-discrimination and keeping the two layers of the Internet digital distribution of content through the Internet. ISJLP 4 (2008), 209. and the Web separate to foster innova- 15. euler, L. Konigsberg Bridge problem Commentarii tion;4 consequently, Internet Science academiae scientiarum Petropolitanae 8 (1741), 128–140. and Web Science remain two distinct 16. Faloutsos, F., Faloutsos, P. and Faloutsos, C. On interdisciplinary areas given they have power-law relationships of the Internet topology. ACM SIGCOMM CCR. Rev. 29, 4 (1999). different goals, those of safeguarding 17. Freeman, L.C. The Development Of Social Network the Internet and the Web, respectively. Analysis. Booksurge LLC, 2004. 18. Frischmann, B.M. Infrastructure. OUP USA, 2012. Network Science explores phenomena 19. Hall, W. and Tiropanis, T. Web evolution and Web Science. Computer Networks 56, 18 (2012), that include, but are not limited to, the 3859–3865. Web or the Internet. 20. Hendler, J. and Berners-Lee, T. From the semantic web to social machines: A research challenge for AI on the However, given the shared pool of . Artificial Intelligence (2009), 1–10. mixed methods and datasets among 21. Hendler, J., Shadbolt, N., Hall, W., Berners-Lee, T. and Weitzner, D. Web Science: an interdisciplinary these three interdisciplinary areas, approach to understanding the Web. Commun. ACM there are compelling benefits for col- 51, 7 (July 2008).

7 communications of the acm | august 2015 | vol. 58 | no. 8