A Two-Stage Clustering Algorithm for Multi-type Relational Data

There are many multi-type relational datasets, the objects in which are multi-type and interrelated. Many clustering methods for this kind of data have been proposed, but because of the complexity of data and relationships, most algorithms have efficiency and scalability problem. To address this difficulty, in this paper a two-stage clustering algorithm for multi-type relational data (TSMRC) has been proposed. Based on the analysis of data and relationships, TSMRC has two stages, which are benefit to improve the efficiency of clustering. To improve the quality of clustering, new similarity measures are proposed, in which attributes and all kinds of relationships are employed. Experimental results on Movie dataset demonstrate the effectiveness of this algorithm.