ICDM 2017 Tutorial


Challenges and Solutions in Group Recommender Systems

Ludovico Boratto (EURECAT, Spain) – ludovico.boratto@acm.org


Abstract. Group recommender systems are designed to provide suggestions in contexts in which people operate in groups. The goal of this tutorial is to provide the ICDM audience with an overview on group recommendation. We will first formally introduce the problem of producing recommendations to groups, then present a survey based on the tasks performed by these systems. We will also analyze challenging topics like their evaluation, and present emerging aspects and techniques in this area. The tutorial will end with a summary that highlights open issues and research challenges.


While recommender systems suggest items that individual users might like, group recommender systems are designed to operate in contexts in which more than one person is involved in the recommendation process [1]. Group recommendation can be naturally adopted in application scenarios that involve groups (e.g., suggest a restaurant to a group of people who want to dine together). In order to filter the items and produce recommendations, Data Mining approaches are largely employed.

Group recommendation has been highlighted as a challenging research area, with the first survey on the topic [1] being placed in the Challenges section of the widely-known book “The Adaptive Web”, and recent research indicating it as a future direction in recommender systems, since it presents numerous open issues and challenges [2]. Indeed, with respect to classic recommender systems, those that operate with groups present several additional aspects that characterize them and cannot be dealt with individual recommender systems. Examples of these challenging aspects (which will be deeply analyzed during the tutorial) are (i) the collection of preferences from both individuals and groups of users, (ii) the modeling of the group preferences, in order to obtain a unique representation of the preferences of a group, starting from the individual ones, (iii) the rating prediction task, which might have to predict ratings straight for a group of users, (iv) the support to a group to help them to achieve a consensus, in order to produce effective recommendations, and (v) the possibility to explain why an item has been recommended to the group, given the individual preferences and the group structure.

Tutorial Outline

The goal of this tutorial is to provide the ICDM audience with an overview on group recommendation. Focusing on the the Data Mining approaches employed to implement the tasks that a group recommender system performs, we will analyze the aspects that characterize these systems, the evaluation methods (evaluating the accuracy for a group is not trivial [2]), the emerging aspects, and the current challenges.

In detail, the outline of the tutorial is the following.

  1. Recommender systems principles
    • Definition and application domains
    • Main classes of systems
  2. Group recommendation introduction
    • Definition and application domains
    • Problem statement
  3. Tasks and state of the art survey
    • Types of groups. In [3], we highlighted that the type of group handled by a system is one of the characterizing aspects of a group recommender system and we provided a classification of the different types of groups, which was also adopted by Carvalho et al. in their WWW’13 paper [4].
    • Preference acquisition. A group recommender system might acquire the preferences by considering only those expressed by the individual users, or by allowing the groups to express them. The interactions among the members in a group can help refining the individual preferences [5].
    • Group modeling [6] is the process adopted to combine the individual preferences in a unique model that represents the group.
    • Rating prediction is the most characterizing aspect in all the types of recommender systems, and also plays an important role when working with groups, since the ratings might be predicted for the individual users or specifically for the groups [1].
    • Help the members to achieve consensus. This task is adopted in order to find an agreement on what should be proposed to the group.
    • Explanation of the recommendations, i.e., the task performed by some of the systems to justify why an item has been suggested to the group.
  4. Evaluation methods
    • Offline methods, which evaluate a system on existing datasets.
    • User surveys that test the effectiveness of a system by asking users to answer questionnaires.
    • Live systems that work in real-world domains, like the social networks.
  5. Emerging aspects and techniques
    • Advanced recommendation techniques applied to group recommendation. The last advances in recommendation techniques, such as matrix factorization and graph- based techniques, have recently been employed in group recommender systems too [7, 8].
    • Fairness in group recommendation. Recently, the problem of how to produce fair recommendations to a group has been analyzed from multiple perspectives [9, 10, 11].
    • Social group recommender systems. The widespread relevance of social media has recently had an impact also on this research area [7,12].
  6. Case study
    • Group recommendation with automatic detection of groups. There are scenarios in which groups do not exist, but the recommendations have to be proposed to groups because of limitations on the number of recommendation lists that can be produced (i.e., it is not possible to suggest a different list of items to each user). Therefore, a clustering of the users specifically designed for recommendation purposes has to be performed [13]. In this section, we will present a case study that shows how the challenges and the tasks presented in Section 3 can be tackled in this specific application scenario (e.g., how to form the groups and how to treat automatically-detected groups for recommendation purposes, how to model the group preferences, how to predict the ratings).
  7. Summary
    • Open issues
    • Research challenges

Target Audience

This tutorial is aimed at anyone interested in the topic of producing recommendation to groups of users, from data mining and machine learning researchers to practitioners from industry. For those not familiar with group recommendation (and recommender systems in general), the tutorial will cover the necessary background material to understand these sys- tems and will provide a state-of-the-art survey on the topic. Additionally, the tutorial aims to offer a new perspective that will be valuable and interesting even for more experienced researchers that work in this domain, by providing the recent advances in this area and by illustrating the current research challenges.


Ludovico Boratto is researcher in the Digital Humanities research group at Eurecat. He previously was a research assistant at the University of Cagliari – Italy. His main research area is Recommender Systems, with special focus on those that work with groups of users and in social environments. His research experience on group recommendation has been recognized with 300+ citations, most of which are related to a state of the art survey on the topic [3]. In 2010 and 2014, he spent 10 months at the Yahoo! Lab in Barcelona as a visiting researcher.


[1] Anthony Jameson and Barry Smyth, “Recommendation to groups,” in The Adaptive Web, Methods and Strategies of Web Personalization, vol. 4321 of Lecture Notes in Computer Science, pp. 596–627. Springer, Berlin, 2007.
[2] Francesco Ricci, “Recommender systems: Models and techniques,” in Encyclopedia of Social Network Analysis and Mining, pp. 1511–1522. Springer New York, 2014.
[3] Ludovico Boratto and Salvatore Carta, “State-of-the-art in group recommendation and new approaches for automatic identification of groups,” in Information Retrieval and Mining in Distributed Environments, vol. 324 of Studies in Computational Intelligence, pp. 1–20. Springer Berlin Heidelberg, 2011.
[4] Lucas Augusto M.C. Carvalho and Hendrik T. Macedo, “Generation of coalition structures to provide proper groups’ formation in group recommender systems,” in Proceedings of the 22Nd International Conference on World Wide Web Companion. 2013, pp. 945–950, International World Wide Web Conferences Steering Committee.
[5] Amra Delic, Julia Neidhardt, Thuy Ngoc Nguyen, Francesco Ricci, Laurens Rook, Hannes Werthner, and Markus Zanker, “Observing group decision making processes,” in Proceedings of the 10th ACM Conference on Recommender Systems, New York, NY, USA, 2016, RecSys ’16, pp. 147–150, ACM.
[6] Judith Masthoff, “Group recommender systems: Combining individual models,” in Recommender Systems Handbook, pp. 677–702. Springer, 2011.
[7] Heung-Nam Kim and Abdulmotaleb El Saddik, “A stochastic approach to group recommendations in social media systems,” Information Systems, vol. 50, no. 0, pp. 76 – 93, 2015.
[8] Quan Yuan, Gao Cong, and Chin-Yew Lin, “Com: A generative model for group recommendation,” in Proceedings of the 20th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. 2014, KDD ’14, pp. 163–172, ACM.
[9] Shuyao Qi, Nikos Mamoulis, Evaggelia Pitoura, and Panayiotis Tsaparas, “Recommending packages to groups,” in IEEE 16th International Conference on Data Mining, ICDM 2016, December 12-15, 2016, Barcelona, Spain, Francesco Bonchi, Josep Domingo-Ferrer, Ricardo A. Baeza-Yates, Zhi-Hua Zhou, and Xindong Wu, Eds. 2016, pp. 449–458, IEEE.
[10] Dimitris Serbos, Shuyao Qi, Nikos Mamoulis, Evaggelia Pitoura, and Panayiotis Tsaparas, “Fairness in package-to-group recommendations,” in Proceedings of the 26th International Conference on World Wide Web, Republic and Canton of Geneva, Switzerland, 2017, WWW ’17, pp. 371–379, International World Wide Web Conferences Steering Committee.
[11] Xiao Lin, Min Zhang, Yongfeng Zhang, and Zhaoquan Gu, “Fairness-aware group recommendation with pareto efficiency,” in Proceedings of the 11th ACM Conference on Recommender Systems (RecSys 2017), August 27 – 31, 2017, Como, Italy. 2017, ACM.
[12] Ingrid Alina Christensen and Silvia Schiaffino, “Social influence in group recommender systems,” Online Information Review, vol. 38, no. 4, pp. 524–542, 2014.
[13] Ludovico Boratto, Salvatore Carta, and Gianni Fenu, “Investigating the role of the rating prediction task in granularity-based group recommender systems and big data scenarios,” Inf. Sci., vol. 378, pp. 424–443, 2017.

credits unica.it | accessibilità Università degli Studi di Cagliari
C.F.: 80019600925 - P.I.: 00443370929
note legali | privacy

Nascondi la toolbar