A Cloud Model-Based Collaborative Filtering Approach for Sparse Data Recommendation Systems

Collaborative filtering is one of the most prominent recommendation methods, particularly excelling in recommending nonlinear resources. However, with the rapid increase in the number of users and items, the scarcity of co-rated items makes user similarity metrics less reliable. To address this, we first predict the ratings of unrated items based on item similarity, thereby increasing the number of co-rated items among users. Then, we employ user similarity measures to identify the nearest neighbor set for each user, and finally, generate recommendations for the target user. The results indicate that this method improves recommendation quality to a certain extent.