Cluster Developing 1-Bit Matrix Completion

1-bit matrix completion plays a significant role in recommender systems. However, conventional 1-bit methods do not consider the interactive clustering nature of customers and products. To tackle the gap, we introduce Group-Specific 1-bit Matrix Completion (GS1MC) by first-time consolidating cluster effects in 1-bit matrix completion tasks. To empower GS1MC even when grouping information is unobtainable, GS1MC is extended as Cluster Developing Matrix Completion (CDMC) by integrating subspace clustering techniques. CDMC can group entities and utilize their cluster effects simultaneously. Experiments show that GS1MC outperforms current 1-bit matrix completion methods and CDMC successfully captures items’ genre features only with sparse binary rating data. Notably, GS1MC provides a new insight to incorporate and evaluate the efficacy of different clustering methods while CDMC can be served as a new tool to explore unrevealed social behavior or market phenomena.