Third Workshop on Recommendation in Complex Scenarios (ComplexRec 2019) Marijn Koolen Toine Bogers Humanities Cluster Department of Communication & Psychology Royal Netherlands Academy of Arts and Sciences Aalborg University Copenhagen Netherlands Denmark marijn.koolen@gmail.com toine@hum.aau.dk Bamshad Mobasher Alexander Tuzhilin School of Computing Stern School of Business DePaul University New York University United States United States mobasher@cs.depaul.edu atuzhili@stern.nyu.edu ABSTRACT variety of constraints on the recommendation task. For example, Over the past decade, recommendation algorithms for ratings pre- standard algorithms typically work with user preferences aggre- diction and item ranking have steadily matured. However, these gated at the item level, but real users may prefer certain features state-of-the-art algorithms are typically applied in relatively straight- of items more than others or attach more weight to those features. forward and static scenarios: given information about a user’s past Furthermore, a user’s interest in an item may vary under different item preferences in isolation, can we predict whether they will like conditions or subject to the peculiarities of the underlying domain. a new item or rank all unseen items based on predicted interest? In Users may want combinations of multiple items, or recommenda- reality, recommendation is often a more complex problem: the eval- tions on the sequence of consumption. Moreover, different users uation of a list of recommended items never takes place in a vacuum, may want different information about items, so beyond ranking and it is often a single step in the user’s more complex background the system needs to decide which information best to display to task or need. The goal of the ComplexRec 2019 workshop is to offer each user. In addition, providing accurate and appropriate recom- an interactive venue for discussing approaches to recommendation mendations in such complex scenarios comes with a whole new in complex scenarios that have no simple one-size-fits-all solution. set of evaluation and validation challenges. Offline datasets do not capture the complexities of online interaction effects related to CCS CONCEPTS different ways of presenting (sets of) recommendations, interaction options and developments of user needs, queries and other interac- • Information systems → Recommender systems. tions throughout sessions. With online evaluation it is a challenge to capture relevant aspects of the user’s current situation, task and KEYWORDS context and to investigate interaction effects between complex sets Complex recommendation of user and data features and interface options. In general, relatively little research has been done on how to 1 INTRODUCTION elicit rich information about these complex background needs or Over the past decade, recommendation algorithms for ratings pre- how to incorporate it into the recommendation process. diction and item ranking have steadily matured, spurred on in part The current generation of recommender systems and algorithms by the success of data mining competitions such as the Netflix Prize, are good at addressing straightforward recommendation scenarios, the 2011 Yahoo! Music KDD Cup, and the RecSys Challenges. Matrix but recommendation under more complex scenarios as described factorization and other latent factor models emerged from these above has not been fully explored. The ComplexRec 2019 work- competitions as the state-of-the-art algorithms to apply in both shop addressed this by providing a interactive venue for discussing existing and new domains. However, these state-of-the-art algo- approaches to recommendation in complex scenarios that have no rithms are typically applied in relatively straightforward and static simple one-size-fits-all solution. scenarios: given information about a user’s past item preferences ComplexRec 2019 was the third edition of the workshop on in isolation, can we predict whether they will like a new item or recommendation in complex scenarios [5, 6]. The first two edi- rank all unseen items based on predicted interest? tions were held at RecSys 20171 and RecSys 20182 . In recent years, In reality, recommendation is often a more complex problem: other workshops have also been organized on topics related to the evaluation of a list of recommended items never takes place our workshop’s focus. Examples include the CARS (Context-aware in a vacuum. It is often a single step in the user’s more complex Recommender Systems) workshop series (2009-2012) organized in underlying task or need and these additional factors often place a conjunction with RecSys [1–4], the CARR (Context-aware Retrieval ComplexRec 2019, 20 September 2019, Copenhagen, Denmark 1 Workshop website and proceedings available at http://complexrec2017.aau.dk/. Copyright ©2019 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).. 2 Workshop website and proceedings available at http://complexrec2018.aau.dk/ ComplexRec 2019, 20 September 2019, Copenhagen, Denmark Marijn Koolen, Toine Bogers, Bamshad Mobasher, and Alexander Tuzhilin and Recommendation) workshop series (2011-2014) organized in 3.1 Keynote conjunction with IUI, WSDM, and ECIR [7, 9–11, 18], as well as the Christoph Trattner described the challenges of providing recom- SCST (Supporting Complex Search Tasks) workshop series (2015, mendations in the domain of food, touching on questions of how 2017) organized in conjunction with ECIR and CHIIR [13, 14]. people make their food choices online, how we can model and predict this behavior, and whether recommender technology can help people change their behaviour towards making healthier food 2 TOPICS AND FORMAT choices by recommendation healthier alternatives to meals they ComplexRec 2019 was organized as an interactive half-day work- like. shop with short paper presentations and a keynote, with the aim of capturing a diverse set of aspects that contribute to complex recommendation scenarios. We therefore invited contributions to the workshop about topics related to complex recommendation, 3.2 Accepted Papers such as: The six accepted papers cover a broad set of complex recommenda- tion scenarios. • Task-based recommendation (Approaches that take the Revina and Rizun [17] presented a concept of a multi-criteria user’s background tasks and needs into account when gen- knowledge-based recommender system that provides decision sup- erating recommendations) port in complex business process scenarios. It utilizes aspects of • Feature-driven recommendation (Techniques for elicit- stylistic patterns, business sentiment and decision-making logic ex- ing, capturing and integrating rich information about user tracted from the unstructured texts, and predicts process complexity preferences for specific product features) and thereby modifies decision support ranging from minimal to • Constraint-based recommendation (Approaches that suc- full automation. cessfully combine state-of-the-art recommendation algo- Doan and Sahebi [12] introduced a hybrid model that jointly rithms with complex knowledge-based or constraint-based models user ratings and reviews across multiple domains, where optimization) knowledge of a user’s preferences and interests in one domain is • Query-driven recommendation (Techniques for eliciting used to recommend items in another domain. It supports decisions and incorporating rich information about the user’s recom- by generating review-like sentences according to user interests and mendation need (e.g., need for accessibility, engagement, item features in more than one domain, with experiments showing socio-cultural values, familiarity, etc.) in addition to the stan- improved transfer of review information. dard user preference information) Collins and Beel [8] provided an analysis of using meta-learning • Interactive recommendation (Techniques for successfully to choose the best recommender algorithm for scholarly article capturing, weighting, and integrating continuous user feed- recommendation per individual session and query document. They back into recommender systems, both in situations of sparse performed both offline and online evaluations, that show that en- and rich user interaction) gagement and click-through rate can be significantly improved by • Context-aware recommendation (Methods for the extrac- selecting the appropriate algorithm based on the user’s currently tion and integration of complex contextual signals for rec- selected document. ommendation) Yang et al. [20] proposed an advice recommender system that • Complex data sources and domains (Approaches to deal- analyses complaint data to recommend web page that contain ad- ing with complex data sources or data sources with unique vice relevant to user dissatisfaction. The system extracts company characteristics in a specific domain or across several domain.) names, complaint topic words and advice topic words from negative • Evaluation & validation (Approaches to the evaluation reviews, and constructs a query from these elements to retrieve and validation of recommendation in complex scenarios) and recommend web pages that offer advice relevant to the review. Murgia et al. [15] discuss the complexities of recommending for young children in the context of education. They identify seven 3 WORKSHOP SUMMARY layers of complexity that recommender systems need to take into The half-day workshop consisted of two slots, with an introduction account, including the different developmental stages that children reviewing the complex scenarios presented in previous ComplexRec can be in and move through at different speeds, the multiple other workshops, as well as six paper presentations and a closing keynote stakeholders in the process like parents and teachers, the impor- presentation by Christoph Trattner about the complex nature of tance of ethics, the fostering learning, providing explanations and online food choices and how this knowledge can be used to build the challenges of assessing what makes a good recommendation. novel food recommender systems [19]. Authors of accepted sub- Naveed and Ziegler [16] focused on the problem of providing missions were invited to give 15-minute presentations. Evaluation feature-driven explanations from hybrid recommenders that the criteria for acceptance included novelty, diversity, significance for user can interact with. A user-feature model is learned from user theory/practice, quality of presentation, and the potential for spark- preferences and item features in the domain of digital cameras, ing interesting discussion at the workshop. All submitted papers which is then used to provide recommendation and explanations. were reviewed by at least three members of the Program Committee. The user can interact with the recommendation by choosing feature- The workshop closed with a brief discussion on future directions based explanations, and by (de)selecting features to use in generat- for research on complex recommendation scenarios. ing new recommendations. Third Workshop on Recommendation in Complex Scenarios (ComplexRec 2019) ComplexRec 2019, 20 September 2019, Copenhagen, Denmark 4 WEBSITE & PROCEEDINGS 10–15. [17] Aleksandra Revina and Nina Rizun. 2019. Multi-Criteria Knowledge-Based The workshop material (list of accepted papers, invited talk, and Recommender System for Decision Support in Complex Business Processes. the workshop schedule) can be found on the ComplexRec work- In Proceedings of the Third Workshop on Recommendation in Complex Scenarios (ComplexRec 2019). 16–22. shop website at http://complexrec2019.aau.dk. The proceedings [18] Alan Said, Ernesto W. De Luca, D. Quercia, and Matthias Böhmer. 2014. 4th Work- are available as a CEUR Workshop Proceedings volume, a link to shop on Context-awareness in Retrieval and Recommendation. In Proceedings of which can be found on the workshop website. A summary of the ECIR ’14. Springer, 802–805. [19] Christoph Trattner. 2019. Online Food Recommendations: A Complex Problem? workshop will appear in SIGIR Forum to increase cross-disciplinary (Keynote). In Proceedings of the Third Workshop on Recommendation in Complex awareness of recommender systems research. In addition, we aim Scenarios (ComplexRec 2019). 4. to explore the possibility of publishing a special journal issue on [20] Liang Yang, Daisuke Kitayama, and Kazutoshi Sumiya. 2019. An Advice Recom- mender System Based on Complaint Data Analysis. In Proceedings of the Third recommendation in complex scenarios, collecting the best authors Workshop on Recommendation in Complex Scenarios (ComplexRec 2019). 35–39. and papers of the 2017-2019 editions of the workshop. REFERENCES [1] Gediminas Adomavicius, Linas Baltrunas, Ernesto William de Luca, Tim Hussein, and Alexander Tuzhilin. 2012. 4th Workshop on Context-aware Recommender Systems (CARS 2012). In Proceedings of RecSys ’12. ACM, New York, NY, USA, 349–350. [2] Gediminas Adomavicius, Linas Baltrunas, Tim Hussein, Francesco Ricci, and Alexander Tuzhilin. 2011. 3rd Workshop on Context-aware Recommender Sys- tems (CARS 2011). In Proceedings of RecSys ’11. ACM, New York, NY, USA, 379–380. [3] Gediminas Adomavicius and Francesco Ricci. 2009. RecSys’09 Workshop 3: Workshop on Context-aware Recommender Systems (CARS-2009). In Proceedings of RecSys ’09. ACM, New York, NY, USA, 423–424. [4] Gediminas Adomavicius, Alexander Tuzhilin, Shlomo Berkovsky, Ernesto W. De Luca, and Alan Said. 2010. Context-awareness in Recommender Systems: Research Workshop and Movie Recommendation Challenge. In Proceedings of RecSys ’10. ACM, New York, NY, USA, 385–386. [5] Toine Bogers, Marijn Koolen, Bamshad Mobasher, Alan Said, and Casper Petersen. 2018. 2nd workshop on recommendation in complex scenarios (ComplexRec 2018). In Proceedings of the 12th ACM Conference on Recommender Systems, RecSys 2018, Vancouver, BC, Canada, October 2-7, 2018, Sole Pera, Michael D. Ekstrand, Xavier Amatriain, and John O’Donovan (Eds.). ACM, 510–511. https://doi.org/10. 1145/3240323.3240332 [6] Toine Bogers, Marijn Koolen, Bamshad Mobasher, Alan Said, and Alexan- der Tuzhilin. 2017. Workshop on Recommendation in Complex Scenarios: (ComplexRec 2017). In Proceedings of the Eleventh ACM Conference on Recom- mender Systems, RecSys 2017, Como, Italy, August 27-31, 2017, Paolo Cremonesi, Francesco Ricci, Shlomo Berkovsky, and Alexander Tuzhilin (Eds.). ACM, 380–381. https://doi.org/10.1145/3109859.3109958 [7] Matthias Böhmer, Ernesto W. De Luca, Alan Said, and Jaime Teevan. 2013. 3rd Workshop on Context-awareness in Retrieval and Recommendation. In Proceed- ings of WSDM ’13. ACM, New York, NY, USA, 789–790. [8] Andrew Collins and Joeran Beel. 2019. Meta-Learning for Scholarly-Article Recommendation. In Proceedings of the Third Workshop on Recommendation in Complex Scenarios (ComplexRec 2019). 29–34. [9] Ernesto William De Luca, Matthias Böhmer, Alan Said, and Ed Chi. 2012. 2nd Workshop on Context-awareness in Retrieval and Recommendation: (CaRR 2012). In Proceedings of IUI ’12. ACM, New York, NY, USA, 409–412. [10] Ernesto William De Luca, Alan Said, Matthias Böhmer, and Florian Michahelles. 2011. Workshop on Context-awareness in Retrieval and Recommendation. In Proceedings of IUI ’11. ACM, New York, NY, USA, 471–472. [11] Ernesto W. De Luca, Alan Said, Fabio Crestani, and David Elsweiler. 2015. 5th Workshop on Context-awareness in Retrieval and Recommendation. In Proceed- ings of ECIR ’15. Springer, 830–833. [12] Thanh-Nam Doan and Shaghayegh Sahebi. 2019. Review-Based Cross-Domain Collaborative Filtering: A Neural Framework. In Proceedings of the Third Workshop on Recommendation in Complex Scenarios (ComplexRec 2019). 23–28. [13] Maria Gäde, Mark Michael Hall, Hugo C. Huurdeman, Jaap Kamps, Marijn Koolen, Mette Skov, Elaine Toms, and David Walsh. 2015. First Workshop on Supporting Complex Search Tasks. In Proceedings of the First International Workshop on Supporting Complex Search Tasks, co-located with ECIR 2015. [14] Marijn Koolen, Jaap Kamps, Toine Bogers, Nicholas J. Belkin, Diane Kelly, and Emine Yilmaz. 2017. Current Research in Supporting Complex Search Tasks. In Proceedings of the Second Workshop on Supporting Complex Search Tasks, co-located with CHIIR 2017. 1–4. [15] Emiliana Murgia, Monica Landoni, Theo Huibers, Jerry Fails, and Maria Soledad Pera. 2019. The Seven Layers of Complexity of Recommender Systems for Children. In Proceedings of the Third Workshop on Recommendation in Complex Scenarios (ComplexRec 2019). 5–9. [16] Sidra Naveed and Jürgen Ziegler. 2019. Feature-Driven Interactive Recommenda- tions and Explanations with Collaborative Filtering Approach. In Proceedings of the Third Workshop on Recommendation in Complex Scenarios (ComplexRec 2019).