A Novel Group Recommendation Algorithm With Collaborative Filtering

Traditional recommender systems are designed to provide suggestions for individuals. However, there are scenarios in which groups of people are in need of decision support. For example, a group of friends want to choose a restaurant to have a dinner or to watch a movie together. In this paper, we propose a novel group recommendation algorithm for providing suggestions to groups. The proposed algorithm can be divided into two steps: the first step is to predict the preference of the unwatched items for each group members, which is a personalized prediction progress; then, it provides the recommendations for the group by aggregating group members' preferences, which mainly concerns the preferences of members who haven't seen the items. Without complex computation, the proposed algorithm can make accurate predictions of each item for group members. We demonstrate our algorithm on a famous dataset called MovieLens and use the recall, the precision metrics and a combination of them to evaluate its performance. The experimental results show that the proposed algorithm can provide high quality group recommendations.