A Survey of Data Mining Method in Heterogeneous Information Networks with Node Importance Evaluation

Abstract

Various kinds of data can be abstracted into information networks, such as the connection between people in social communication, the publishing and published relationship between authors and papers in academic network, etc. Heterogeneous information networks composed of various types of nodes and edges contain more comprehensive and rich structural and semantic information. The node importance evaluation in heterogeneous information networks can not only be based on different types of heterogeneous information in the network, but also make the evaluated key nodes more meaningful and valuable. In recent years, the node importance evaluation in heterogeneous information networks has been widely concerned by academia and industry, and has become an important research topic of network analysis. However, at present, there is still a lack of work to comprehensively sort out the existing results, and it is difficult for relevant researchers to systematically understand the latest research progress and choose appropriate data mining methods in practical applications. Therefore, this paper comprehensively summarizes the data mining methods used in the evaluation of key nodes of heterogeneous information networks. This paper includes: the basic concept of heterogeneous information network; the latest research progress of data mining methods which used for node importance evaluation; the latest research progress of node importance evaluation of heterogeneous information network. Then we prospect the future development direction of node importance evaluation.