异构信息网络上相似性搜索的研究.pdf

异构信息网络上的相似性搜索研究 万方数据 Abstract Abstract In recent years, heterogeneous information network analysis becomes a very popular and novel research direction in the field of data mining, attracting more and more attention of experts and scholars. In the real world, systems of different areas often exist multiple types of objects and different types of links between the different types of objects and can be modeled as a heterogeneous information network. Similarity search is a typical job in heterogeneous information network analysis, By researching similarity search in heterogeneous information network we can effectively mine the rich semantic and hidden knowledge of networks. This paper analyzes the traditional feature-based similarity measurement and the current link-based similarity search algorithms and their shortage under homogeneous and heterogeneous information network. Aiming at the problem in PathSim similarity search algorithm that results generate influence bias and area bias, this paper puts forward the concept and calculation method of object influence factor in heterogeneous information network, and on this basis proposes a similarity search algorithm APSim which synthesize influence factor and based on the meta path. The algorithm leads meta path-based similarity search in heterogeneous information network more practical significative by adding the