论文信息 - Opinion-Based Collaborative Filtering to Solve Popularity Bias in Recommender Systems

Opinion-Based Collaborative Filtering to Solve Popularity Bias in Recommender Systems

Existing recommender systems suffer from a popularity bias problem. Popular items are always recommended to users regardless whether they are related to users’ preferences. In this paper, we propose an opinion-based collaborative filtering by introducing weighting functions to adjust the influence of popular items. Based on conventional user-based collaborative filtering, the weighting functions are used in measuring users’ similarities so that the effect of popular items is decreased with similar opinions and increased with dissimilar ones. Experiments verify the effectiveness of our proposed approach.

Wei Chen | Zhendong Niu | Xiangyu Zhao

[1] R. Merton. The Matthew Effect in Science , 1968, Science.

[2] David Heckerman,et al. Empirical Analysis of Predictive Algorithms for Collaborative Filtering , 1998, UAI.

[3] Ali R. Hurson,et al. Ontology-Based Recommendation Algorithms for Personalized Education , 2012, DEXA.

[4] Sun Park,et al. Novel Recommendation Based on Personal Popularity Tendency , 2011, 2011 IEEE 11th International Conference on Data Mining.

[5] Gediminas Adomavicius,et al. Improving Aggregate Recommendation Diversity Using Ranking-Based Techniques , 2012, IEEE Transactions on Knowledge and Data Engineering.

[6] Jaana Kekäläinen,et al. Cumulated gain-based evaluation of IR techniques , 2002, TOIS.

[7] Cheng Zeng,et al. A New Approach for Date Sharing and Recommendation in Social Web , 2012, DEXA.

[8] Hang Li,et al. Hybrid Recommendation Models for Binary User Preference Prediction Problem , 2012, KDD Cup.

[9] Luo Si,et al. An automatic weighting scheme for collaborative filtering , 2004, SIGIR '04.

[10] Yi Li,et al. A hybrid recommendation algorithm adapted in e-learning environments , 2012, World Wide Web.

[11] Wenfei Fan,et al. Keys with Upward Wildcards for XML , 2001, DEXA.