%A Chen Meimei,Xue Kangjie %T Personalized Recommendation Algorithm of Multi-faceted Trust Tensor Based on Tag Clustering %0 Journal Article %D 2017 %J Data Analysis and Knowledge Discovery %R 10.11925/infotech.2096-3467.2017.05.11 %P 94-101 %V 1 %N 5 %U {https://manu44.magtech.com.cn/Jwk_infotech_wk3/CN/abstract/article_4370.shtml} %8 2017-05-25 %X

[Objective] This paper aims to solve the low accuracy issue facing personalized recommendation algorithm of multi-faceted trust tensor based on tag clustering. [Methods] First, we proposed a new method to calculate multi-faceted trust based on tag clusters. Then, we introduced the TF-IDF and Pearson similarity to indicate strength of inter-cluster and intra-cluster trust. Finally, we built recommendation mechanism based on tensor decomposition to reflect the trust intensity from different facets. [Results] We examined the new algorithm with the Last.fm dataset. The precision, recall and F1 measures were better than traditional methods. Among them, the F1 measure was increased by 2.29% on average. [Limitations] Our new algorithm needs to be examined with datasets from Weibo or Twitter. [Conclusions] The proposed algorithm could effectively increase the accuracy of recommendation by defining and quantifying trust relationship among users. It improves the user experience of social network systems.