Clustering for heterogeneous information networks with extended star-structure

• Published in: Data mining and knowledge discovery Vol. 33; no. 4; pp. 1059 - 1087
• Main Authors: Mei, Jian-Ping, Lv, Huajiang, Yang, Lianghuai, Li, Yanjun
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.07.2019; Springer Nature B.V
• Subjects: Artificial Intelligence; Chemistry and Earth Sciences; Clustering; Computer Science; Data Mining and Knowledge Discovery; Information Storage and Retrieval; Links; Network analysis; Networks; Physics; Statistics for Engineering; Clustering; Multi-type relational data; Heterogeneous information network
• ISSN: 1384-5810; 1573-756X
• DOI: 10.1007/s10618-019-00626-2

Clustering of objects in a heterogeneous information network, where different types of objects are linked to each other, is an important problem in heterogeneous information network analysis. Several existing clustering approaches deal with star-structured information networks with different central-attribute relations. In real applications, homogeneous links between central objects may also be available and useful for clustering. In this paper, we propose a new approach called CluEstar for clustering of network with an extended star-structure (E-Star), which extends the classic star-structure by further including central–central relation, i.e., links between objects of the central type. In CluEstar, all objects have a ranking with respect to each cluster to reflect their within-cluster representativeness and determine the clusters of objects that they linked to. A novel objective function is proposed for clustering of E-Star network by formulating both central-attribute and central–central links in an efficient way. Results of extensive experimental studies with benchmark data sets show that the proposed approach is more favorable than existing ones for clustering of E-Star networks with high quality and good efficiency.