Group Formation Based on Crowdsourced Top-k Recommendation

Abstract

There has been significant recent interest in the area of group recommendations, where, given groups of users of a recommender system, one wants to recommend $top-k$ items to a group to achieve an object such as minimizing the maximum disagreement between group members, according to a chosen semantics of group satisfaction. We consider the complementary problem of how to form groups such that the users in the formed groups are as even satisfied with the suggested $top-k$ recommendations as possible. Thanks to emerging crowdsourcing platforms, e.g., Amazon Mechanical Turk and CrowdFlower, we have easy access to workers’ $top-N$ recommendation. Here dealing with the ranking data is a big challenge, as quite few methods to solve this issue. We assume that the recommendations will be generated according to minimize the maximum disagreement between group users utilizing the kendall distance. Rather than assuming groups are given, or rely on ad hoc group formation dynamics, our framework allows a strategic approach for forming groups of users in order to minimize the maximum disagreement. Furthermore, we develop efficient algorithms for group formation under the minmax object. We validate our results and demonstrate the scalability and effectiveness of our group formation algorithms on both real and synthetic data sets.