Ensuring Fairness in Group Recommendations by Rank-Sensitive Balancing of Relevance

Abstract

For group recommendations, one objective is to recommend an ordered set of items, a top-, to a group such that each individual recommendation is relevant for everyone. A common way to do this is to select items on which the group can agree, using so-called 'aggregation strategies'. One weakness of these aggregation strategies is that they select items independently of each other. They therefore cannot guarantee properties such as fairness, that apply to the set of recommendations as a whole.In this paper, we give a definition of fairness that 'balances' the relevance of the recommended items across the group members in a rank-sensitive way. Informally, an ordered set of recommended items is considered fair to a group if the relevance of the items in the top-is balanced across the group members for each prefix of the top-. In other words, the first item in the top-should, as far as possible, balance the interests of all group members; the first two items taken together must do the same; also the first three; and so on up to . In this paper, we formalize this notion of rank-sensitive balance and provide a greedy algorithm (GFAR) for finding a topset of group recommendations that satisfies our definition.We compare the performance of GFAR to five approaches from the literature on two datasets, one from each of the movie and music domains. We evaluate performance for 42 different configurations (two datasets, seven different group sizes, three different group types) and for ten evaluation metrics. We find that GFAR performs significantly better than all other algorithms around 43% of the time; in only 10% of cases are there algorithms that are significantly better than GFAR. Furthermore, GFAR performs particularly well in the most difficult cases, where groups are large and interests within the group diverge. We attribute GFAR's success both to its rank-sensitivity and its way of balancing relevance. Current methods do not define fairness in a rank-sensitive way (although some achieve a degree of rank-sensitivity through the use of greedy algorithms) and none define balance in the way that we do.

Rank-sensitive proportional aggregations in dynamic recommendation scenarios

“…In this paper, we evaluate one possible use-case for the framework, namely proportional aggregation of multiple base recommending algorithms. However, let us already now highlight that the framework or its parts may be useful in other areas of RS research as well, e.g., group recommendation problem (Kaya et al 2020), proportional representation of items' sub-components (Starychfojtu and Peska 2020) or calibrating various quality criteria (novelty, diversity, relevance) of recommendations.…”

“…One of the well known disadvantages of item-wise aggregations is the risk of a systematic bias against some of the aggregated parties (i.e., recommenders in our case) (Kaya et al 2020;Serbos et al 2017;Xiao et al 2017). For instance, consider the following example with three recommender systems R 1 , R 2 and R 3 that propose items as shown in Table 1.…”

“…Several group RS focused on the list-wise aggregations (Xiao et al 2017;Sacharidis 2019;Serbos et al 2017;Kaya et al 2020), where some concept of fairness among group members is introduced and maintained in the final list of recommendations. An early example is FAI algorithm (Masthoff 2011), which regularly switches between users and selects the best remaining item of the current user.…”

“…One further enhancement to the above-described methodology was proposed by Kaya et al (2020). Instead of focusing on the overall fairness of the whole list, they introduced a rank-sensitive fairness property of the recommendations and proposed a greedy optimization algorithm (GFAR) that maintains this property.…”

Sequential group recommendations based on satisfaction and disagreement scores

“…They construct the group recommendation list by considering sets of N-level Pareto optimal items. Kaya et al (2020) proposes the notion of rank-sensitive balance in order to achieve fairness during the aggregation phase. All these works, for achieving fairness, consider one instance of group recommendations and do not take into account the sequential group recommendation problem, as we do in our work.…”