Title: Group Recommender Systems with Group Dynamic Learning
Abstract:
In the field of electronic commerce, recommender system is a powerful tool that helps the consumers to identify potentially desired products. Different approaches has evolved in attempt to achieve accurate results, most notably content-filtering and collaborative-filtering. While previous research has primarily targeted at the development of recommender systems for single users, this research focuses on applying the recommender system concept to a group of users instead.
The research proposes two novel mechanisms - profile merger and recommendation merger - to produce recommendation for groups of users, while relying on the advanced algorithms developed for the collaborative filtering approach. In addition, the proposed algorithm includes the capability to model and adapt the influence of group dynamics that exist in any arbitrary groups. The primary objective of the algorithm is the accuracy of the produced recommendations for the entire group, and is evaluated by gathering user feedbacks and measured through a multitue of metrics.