A Split-Merging K-Means Clustering Algorithm with User Attributes for Collaborative Filtering

Collaborative Filtering (CF) stands out as a robust recommendation method that can significantly improve the accuracy of the recommendation system. Nevertheless, conventional CF algorithms face challenges such as cold start and data sparsity. This study presents an approach that combines a split-merging clustering algorithm with user attributes to improve CF performance. Initially, the Split-Merging Clustering (SMKC) algorithm is utilized to cluster users based on their rating vectors, resulting in SMKC. Subsequently, the similarity of the user is calculated by integrating attributes such as age, gender, and occupation to identify the nearest neighbors. Finally, item evaluations are predicted based on user similarities within the nearest-neighbor set to generate the list of recommendations. Experimental findings demonstrate that the proposed method outperforms traditional approaches in generating recommendation lists, addressing issues related to data sparsity and the inherent cold start in the data.