Research On Key Technologies Of Meta Path Mining In Social Networks

With the continuous development and improvement of social network functions,people gradually start to join different social networks for daily communication.All social networks are trying to increase the number of users through the analysis of various user entity relationships by trying to meet users' social needs.Social network is generally by providing a wide range of different social services to attract users,users through social networks to provide services to build personal social circles,so through the study of a variety of different network users,such as meta path,anchor link relations and community associations,etc.,can effectively improve the analysis of the social network and service quality.This paper mainly studies the key technologies of meta-path mining in social networks from two aspects: meta-path mining technology based on feature optimization and path quality evaluation algorithm based on metapath mining.Firstly,in order to solve the problem that the meta-path features in meta-path mining are simple and not ful y adapted to large-scale social networks,this paper proposes a feature optimization algorithm for meta-paths.Most of the existing meta-path features in meta-path mining are only applicable to homogeneous networks,and a few meta-path features are applicable to heterogeneous networks,but these meta-path features cannot be used well for information mining in large-scale heterogeneous social networks and meta path mining.To solve the above problems,from the perspective of feature optimization,the feature optimization algorithm of meta-paths proposes a variety of single-optimized path features and through the compound optimization of multiple different meta-path features,it is more outstanding in social network meta-path mining Meta path characteristics.Experimental verification shows that the meta-path feature optimization algorithm can obtain better meta-path features and has obvious advantages in meta-path mining of heterogeneous social networks.Secondly,Secondly,in order to solve the problem of meta-path mining in large-scale social networks,this paper proposes a path quality assessment algorithm for meta-paths.Most existing meta path mining methods use meta paths in the network indiscriminately,without taking account the difference in the number and importance of different types of meta paths in different social networks,which will make it impossible to evaluate and meta path.Mining the problem of inaccurate results.In order to solve the above problems,the path quality evaluation algorithm of the meta path starts from the perspective of path quality evaluation,calculates the similarity between entities for entities on different meta paths,and evaluates the path quality based on the meta similarity of the entities.Obtain the evaluation of social network meta-path mining as more valuable meta-paths and their meta-path characteristics.Experimental verification shows that the meta-path path quality evaluation algorithm can obtain better metapaths and their characteristics,and has obvious advantages in meta-path mining of heterogeneous social networks.