Relevance Measure in Large-Scale Heterogeneous Networks

Abstract

Recently, there is a surge of heterogeneous information network analysis, where network includes multiple types of objects or links. Many data mining tasks have been studied on it, among which similarity measure is a basic and important function. Several similarity measures have been proposed in heterogeneous information network. However, they suffer from high computation and memory demand. In this paper, we propose a novel measure, called AvgSim, which can measure similarity of same or different-typed object pairs in a uniform framework and has some good properties. AvgSim value of two objects is evaluated through two random walk processes along the given meta-path and the reverse meta-path, respectively. In addition, we implement AvgSim using MapReduce parallel model in order to enable the application in large-scale networks. Experiments on real data sets verify the effectiveness and efficiency of AvgSim.