Workshop proceedings ComplexRec 2018 Second Workshop on Recommendation in Complex Scenarios October 7, 2018 RecSys 2018, Vancouver, Canada Edited by Casper Petersen, Toine Bogers, Marijn Koolen, Bamshad Mobasher, Alan Said Contents 1 Preface 3 2 Program Committee 4 3 Workshop Overview 5 4 Accepted Papers 8 4.1 Recommendations for sports games to bet on . 8 4.2 Finding Your Home: Large-Scale Recommendation in a Vibrant Marketplace . 13 4.3 User and Context Aware Composite Item Recommendation . 18 4.4 Recommendations for repeat consumption products considering users’ tendency to repeat . 22 4.5 Recommending Running Routes: Framework and Demonstrator . 26 4.6 Time-aware Personalized Popularity in top-N Recommendation . 30 4.7 Retrieving and Recommending for the Classroom: Stakeholders Objectives Resources and Users 34 2 ComplexRec 2018 Preface Preface This volume contains the papers presented at the RecSys 2018 workshop Recommendation in Complex Scenarios (ComplexRec 2018) held on October 7, 2018 at the Parq Vancouver, in Vancouver, Canada. State-of-the-art recommendation algorithms are typically applied in relatively straightforward and static scenarios: given information about a user’s past item preferences in isolation, can we predict whether they will like a new item or rank all unseen items based on predicted interest? In reality, recommendation is often a more complex problem: the evaluation of a list of recommended items never takes place in a vacuum, and it is often a single step in the user’s more complex background task or need. These background needs can often place a variety of constraints on which recommendations are interesting to the user and when they are appropriate. However, relatively little research has been done on how to elicit rich information about these complex background needs or how to incorporate it into the recommendation process. Furthermore, while state-of-the-art algorithms typically work with user preferences aggregated at the item level, real users may prefer some of an item’s features more than others or attach more weight in general to certain features. Finally, providing accurate and appropriate recommendations in such complex scenarios comes with a whole new set of evaluation and validation challenges. The current generation of recommender systems and algorithms are good at addressing straight- forward recommendation scenarios, but the more complex scenarios as described above have been underserved. The ComplexRec 2018 workshop aims to address this by providing an interactive venue for discussing approaches to recommendation in complex scenarios that have no simple one-size-fits-all solution. The workshop program contains a set of position and research papers covering many complex as- pects of recommendation in various scenarios. There were 14 submissions. Each submission was reviewed by at least 3 program committee members. The committee decided to accept 7 papers (ac- ceptance rate 50%). We thank the program committee members for their timely and constructive reviews. We gratefully acknowledge the support of EasyChair for organizing paper submission and reviewing and producing the proceedings. August 28, 2018 Casper Petersen Copenhagen Toine Bogers Marijn Koolen Bamshad Mobasher Alan Said 3 ComplexRec 2018 Program Committee Program Committee Haggai Roitman IBM Research Haifa Jaap Kamps University of Amsterdam Soude Fazeli Open University of Netherlands Nafiseh Shabib Norwegian University of Science and Technology Robin Burke DePaul University Paolo Cremonesi Politecnico di Milano Marco De Gemmis Dipartimento di Informatica - University of Bari Fedelucio Narducci University of Bari Aldo Moro Cristina Gena Department of Computer Science, University of Torino Tommaso Di Noia Politecnico di Bari Fabio Gasparetti Artificial Intelligence Laboratory - ROMA TRE University Peter Dolog Aalborg University Cataldo Musto Dipartimento di Informatica - University of Bari Federica Cena Department of Computer Science, University of Torino Oren Sar Shalom Bar Ilan University Ludovico Boratto Eurecat Markus Schedl Johannes Kepler University Panagiotis Adamopoulos Emory University Frank Hopfgartner The University of Sheffield Juan F. Huete University of Granada Ivan` Cantador Universidad Autonoma´ de Madrid Ernesto William De Luca Georg-Eckert-Institute - Leibniz-Institute for international Textbook Research 4 2nd Workshop on Recommendation in Complex Scenarios (ComplexRec 2018) Toine Bogers Marijn Koolen Bamshad Mobasher Department of Communication & Huygens ING, Royal Netherlands School of Computing Psychology Academy of Arts and Sciences DePaul University Aalborg University Copenhagen Netherlands United States Denmark [email protected] [email protected] [email protected] Alan Said Casper Petersen University of Skövde Sampension Sweden Denmark [email protected] [email protected] ABSTRACT both existing and new domains. However, these state-of-the-art Over the past decade, recommendation algorithms for ratings pre- algorithms are typically applied in relatively straightforward and diction and item ranking have steadily matured. However, these static scenarios: given information about a user’s past item pref- state-of-the-art algorithms are typically applied in relatively straight- erences in isolation, can we predict whether they will like a new forward scenarios. In reality, recommendation is often a more item or rank all unseen items based on predicted interests? complex problem: it is usually just a single step in the user’s more In reality, recommendation is often a more complex problem: complex background need. These background needs can often place the evaluation of a list of recommended items never takes place in a a variety of constraints on which recommendations are interesting vacuum, and it is often only a single step in the user’s more complex to the user and when they are appropriate. However, relatively little background task or need. These background needs can often place research has been done on these complex recommendation scenar- a variety of constraints on which recommendations are interesting ios. The ComplexRec 2018 workshop addresses this by providing to the user and when they are appropriate. However, relatively little an interactive venue for discussing approaches to recommendation research has been done on how to elicit rich information about in complex scenarios that have no simple one-size-ts-all solution. these complex background needs or how to incorporate it into the recommendation process. Furthermore, while state-of-the-art KEYWORDS algorithms typically work with user preferences aggregated at the item level, real users may prefer some of an item’s features more Complex recommendation than others or attach more weight in general to certain features. ACM Reference format: Finally, providing accurate and appropriate recommendations in Toine Bogers, Marijn Koolen, Bamshad Mobasher, Alan Said, and Casper such complex scenarios comes with a whole new set of evaluation Petersen. 2018. 2nd Workshop on Recommendation in Complex Scenarios and validation challenges. (ComplexRec 2018). In Proceedings of Twelfth ACM Conference on Recom- The current generation of recommender systems and algorithms mender Systems, Vancouver, BC, Canada, October 2–7, 2018 (RecSys ’18), 3 pages. are good at addressing straightforward recommendation scenar- DOI: 10.1145/3240323.3240332 ios, yet more complex scenarios as described above have been underserved. The ComplexRec 2018 workshop addresses this by providing an interactive venue for discussing approaches to 1 INTRODUCTION recommendation in complex scenarios that have no simple one- Over the past decade, recommendation algorithms for ratings pre- size-ts-all solution. It is the second edition of this workshop, diction and item ranking have steadily matured, spurred on in part after a successful rst edition in 2017 [5]. In addition to this rst by the success of data mining competitions such as the Netix edition, other workshops have also been organized on related top- Prize, the 2011 Yahoo! Music KDD Cup, and the RecSys Challenges. ics in recent years. Examples include the CARS (Context-aware Matrix factorization and other latent factor models emerged from Recommender Systems) workshop series (2009-2012) organized in these competitions as the state-of-the-art algorithms to apply in conjunction with RecSys [1–4], the CARR (Context-aware Retrieval and Recommendation) workshop series (2011-2015) organized in Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed conjunction with IUI, WSDM, and ECIR [6–9, 12], as well as the for prot or commercial advantage and that copies bear this notice and the full citation SCST (Supporting Complex Search Tasks) workshop series (2015, on the rst page. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, 2017) organized in conjunction with ECIR and CHIIR [10, 11]. to post on servers or to redistribute to lists, requires prior specic permission and/or a fee. Request permissions from [email protected]. RecSys ’18, Vancouver, BC, Canada © 2018 ACM. 978-1-4503-5901-6/18/10...$15.00 DOI: 10.1145/3240323.3240332 5 2 FORMAT & TOPICS a prototype recommendation tool based on dierent classication ComplexRec 2018 will be organized as an interactive, half-day work- models. Their results show e.g. that the