Structural Hole Mining Algorithm Based On Meta-Path Weighting In Heterogeneous Networks

With development of society,social networks have become the network that covers the widest range of users and has the greatest impact on communication.The analysis of social networks is a hotspot in the field of data mining research and has attracted the attention of more and more researchers.In an social network,two individuals that cannot be directly connected can only be connected by a third party.Those nodes in the third position are called structural holes.Nodes at structural hole locations are able to capture"information benefits"and"control bene:fits"and thus have more competitive advantages than members in other locations in the network.This part of the node plays an extremely important role in the dissemination and influence of group information in the network.Therefore,the identification of structural holes is an important research content of social network analysis.Regarding the identification of structural holes in social networks,most of the current research is mainly carried out in homogeneous networks containing one type of nodes and edges.However,real-life networks are often heterogeneous.They contain many types of nodes and edges,so they contain richer semantic information.Although there have been some researches on the mining of structural holes in heterogeneous networks,there is not much involvement of meta-paths and weighting problems.When constructing structural holes in heterogeneous networks,the processing of different meta-paths and different weights is required.A deeper study.Due to the particularity of heterogeneous networks,the traditional structural hole mining algorithm can not be directly applied to heterogeneous information networks.A weighting method for meta-paths in heterogeneous networks is proposed for structural hole mining.The main work of this paper is as follows:First,calculate the degree of association between nodes.Exploring the more comprehensive meta-path semantic information between nodes in a heterogeneous network,searching for the e-go path through the breadth-first algorithm,and assigning different weights to the edges on the meta-path,and then performing a weighted instance path number matrix between the target nodes.The summation forms an associative matrix.Secondly,a structural hole mining algorithm based on meta-path weighting is proposed.Combining the association matrix formed between the target types with the structural hole mining method based on the shortest path increment,the heterogeneous network rich semantic information is used to mine the structure.Finally,the design and implementation of the experiment are carried out,and compared with the traditional structural hole mining algorithm on the real data set.The experimental results show that the proposed structural hole identification method can make a more comprehensive search for the structural hole nodes in the network.It also demonstrates the effectiveness of the structural hole mining algorithm for weighted weights in heterogeneous networks.