Group Recommendation

When people make decisions together, diverse and sometimes conflicting preferences make group recommendation more than a simple aggregation problem. We study methods that balance individual interests, social relationships, and collective satisfaction rather than optimizing for a single user.

We explore multi-feedback learning, tensor-based models, and combinatorial optimization to represent different preference signals and reconcile competing interests. These approaches jointly consider users, items, feedback, and group formation to recommend choices that work for the group without overlooking individual members.

Our work applies these ideas to shared virtual-reality shopping and live multi-streaming, while extending toward emerging scenarios in which people interact, negotiate preferences, and make decisions together.