基于信任关系和时间衰减效应的协同过滤推荐算法

Collaborative Filtering Recommendation Algorithm Based on Trust Relationship and Time Decay Effect

经典的协同过滤推荐模型借助于用户评分进行商品推荐,而数据稀疏性往往导致推荐准确率不高,为了缓解该问题,在传统算法基础上构建了基于信任关系优化用户评分相似度和基于时间衰减效应优化用户兴趣相似度的协同过滤推荐算法,融合这两个维度的相似性生成用户间的最终相似性,以此寻找目标用户的相似用户群,进而实施商品推荐。最后仿真实验表明该方法相较于对照算法在MAE、Precision、Recall、Coverage指标上能获得更好的推荐效果。

At present,classic collaborative filtering recommendation models rely on user ratings for product recommendations,and the sparseness of the data often leads to low recommendation accuracy.In order to alleviate this problem,this paper builds a collaborative filtering recommendation algorithm based on traditional algorithms to optimize user score similarity based on trust relationships and to optimize user interest similarity based on time decay effects.And merging the similarities of these two dimensions generates the final similarity between users,so as to find the similar user group of the target user and to implement product recommendation.Finally,simulation experiments show that the proposed method can obtain better recommendation results on the MAE,Precision,Recall,and Coverage indicators compared with the control algorithm.