Algorithm and Information Colloquium (AIC)

Title:　　　  Tags Meet Ratings: Improving Collaborative Filtering with Tag-Based Neighborhood Method

Speaker:　  Zhe Wang 王哲

Time:　　　　  16:30 – 17:30, May 5, 2010, Wednesday

Place:　　　　 Lecture Room, Level 3, Building No. 5, Institute of Software

Tea time:　　 16:00 – 16:30, Common Room, Level 3, Building No. 5

Abstract:

Collaborative Filtering (CF) is a method for personalized recommendation. The sparsity of rating data poses a heavy influence on the quality of CF's recommendation. Meanwhile, there is more and more tag information generated by online users that implies their preferences. Exploiting these tag data is a promising means to alleviate the sparsity problem. Although the intention is straight-forward, there's no existed solution that makes full use of tags to improve the item recommendation quality of traditional rating-based collaborative Filtering approaches. Here we propose a novel approach to fuse a tag-based neighborhood method into the traditional rating-based CF. Tag-based neighborhood method is employed to find similar users and items. This neighborhood information helps the sequent CF procedure produce higher quality recommendations. The experiments show that our approach outperforms the state-of-the-art ones.

Biography:

Zhe Wang is currently a master degree student in National Engineering Research Center of Fundamental Software, Institute of Software, Chinese Academy of Sciences. Before this, he earned his B.S. in Computer Science degree in Beijing Institute of Technology. Up to now, he has some technical publications in the areas of recommender system and personalized information retrieval. In particular, Mr. Wang is interested in Bayesian inference and matrix computation.