第 卷 第 期 沈 阳 理 工 大 学 学 报 Vo l No
40 6 mounts of dirty data in an explosive way. In order to get higher quality data without being
affected by these dirty data people begin to focuson Recommendation Systems. Traditional
ꎬ
recommendation algorithms such as Collaborative Filtering CF rely too much on the in ̄
ꎬ ( )ꎬ
teractive information between users and items so the problems of data sparsity and cold start
ꎬ
have become difficult problems to be solved by the academia all the time. Recently Knowl ̄
ꎬ
edge Graph has been widely used in recommendation systems because of its trituple′s intelli ̄
gibility and rich semantic information. For these reasons we propose a recommendation al ̄
ꎬ
gorithm model which is based on knowledge embedding model TransE and Graph neural
ꎬ
network. Higher ̄order representation of users and items can be extracted by using the Graph
neural network. Knowledge graph embedding algorithm model can extract the characteristic
informa
基于知识图谱的推荐算法研究 来自淘豆网m.daumloan.com转载请标明出处.