Using Cloud Model for Default Voting in Collaborative Filtering
Kuo-Cheng Tseng, Chein-Shung Hwang, YiChing Su - · Journal of Convergence Information Technology · 2011
Many personalized recommender systems have been developed for e-commerce applications. Most recommender systems implement a neighborhood-based CF approach in which the neighborhood generation is based on the pair-wise user similarities. However, the similarity computation is not reliable when the ratings data is sparse. Default voting is a common technique to ameliorate the sparsity problem. In this study, we propose a new default voting scheme using the cloud model. The cloud model represents the user’s global preference that is computed from users’ past ratings. The experimental results show that the new default voting scheme can improve the performance of traditional neighborhood-based CF approach with sparse data.