基于用户相似性传递的协同过滤推荐方法

Collaborative Filtering Recommendation Method Based on User Similarity Propagation

随着Web服务量日益剧增,如何推荐给用户符合其期望的服务尤为重要.协同过滤是服务推荐系统中应用较广泛的算法,但协同过滤往往存在数据稀疏问题.针对该问题,文中提出了一种基于用户相似性传递的协同过滤推荐算法,通过改进的皮尔逊相关系数计算出用户间的相似性,利用用户相似性构建相似性网络,并根据用户等级、最短路径和六度分隔理论进行用户间的相似性传递,以此增加目标用户的相似用户的数量,缓解了数据稀疏问题,然后根据传递填充过的用户相似网络进行服务的评分预测.最后以美团网的真实数据进行了实验,实验结果表明文中提出方法一定程度上可以提高推荐结果的准确性.

With the increasing growth of Web services,how to recommend services to users that meet their expectations is especially important.Collaborative filtering is a widely used algorithm in service recommendation systems,but it often has sparse data.For this problem,this paper proposed an improved collaborative filtering recommendation algorithm based on user similarity propagation.The similarity between users was calculated by the improved Pearson correlation coefficient,and the similarity network was further constructed by user similarity.In order to alleviate the problem of data sparseness by increasing the number of similar users of the target user,we transfered the similarity among users according to the user level,shortest path and six degree separation theory.Then we predicted the service score by the propagated user similarity network.Finally,we conducted experiments on the real datasets from meituan network.The experimental results show that the proposed method can improve the accuracy of the recommendation results to a certain extent.