A group recommendation method for differentiated member preferences based on user-item bipartite graph conditional walk
Compared with individual recommendation,group recommendation faces the problem of inaccurate expression of member preferences and group preferences.In the case of limited data,in order to accurately express member and group prefer-ences,a differentiated member preference group recommendation method based on bipartite graph conditional walk is proposed.First,conditional walks are carried out on the user-item bipartite graph to form conditional paths,and node information is aggre-gated to form member preferences.And for different items,a differentiated attention mechanism is used to assign different weights to members in the group,aggregate overall preferences,calculate the group's scores for recommended items and complete the rec-ommendation.This experiment is tested on a real dataset to verify the effectiveness of the proposed method.
group recommendationuser-item bipartite graphconditional walkpreference differencedynamic aggregation
万方数据
