Research on WEB user behavior mining of personalized recommendation
Xufang Li1, 2
COMPUTER MODELLING & NEW TECHNOLOGIES 2014 18(12B) 50-53
1 School of Economics and Management, Tongji University, Shanghai, 200092, China
2 School of Management, Shanghai University of Engineering Science, Shanghai, 201620, China
Personalized recommendation directly decides the result set pushed to users and affects the quality of personalized information service. And analysis of user behavior is the key to realize personalized recommendation. The paper studies the user behavior mining based on content and VecPat-tree. When mining based on content, compound word judgment is joined in segmentation process, and the concept of keyword position factor is added to keyword weight calculation. When mining based on VecPat-tree, the paper proposed the algorithm based on VecPat-tree to process user behavior mining. The algorithm based on VecPat-tree uses the strategy of binary tree growth to avoid unnecessary projected database and effectively distinguish distribution and partial support. The paper simulated 193000 browse records of users in the experimental database to compare PrefixSpan algorithm and the algorithm based on VecPat-tree in many aspects, such as running time. And the experimental results show that the algorithm based on VecPat-tree can be more effective than PrefixSpan algorithm in achieving personalized recommendation to improve the quality of personalized information service.