Improving the diversity of user-based Top-N recommendation by Cloud Model
Bing Wang, Zhaowen Tao, Jun Hu · 2010
Recommender system is one of the most effective technologies to deal with information overload, which has been used in a lot of business systems. Historically, many recommender systems take much focus on prediction accuracy. However, despite their pretty accuracy, they may not be useful to users. A user's preference is full of uncertainty, including randomness and fuzziness. Unfortunately, a fixed Top-N recommendation list certainly can not describe this kinds of uncertainty, which has leaded a decline of user satisfaction. Cloud Model is a powerful tool to describe uncertainty of knowledge. In this paper, we use Cloud Model to present user's preference and propose a improved user-based Top-N recommendation algorithm. Our experimental evaluation show that our proposed algorithm can improve the diversity of recommendation list compared with the typical user-based collaborative filtering.