Research On Relation Inference And Prediction In Heterogeneous Netwok

With the rapid development of Internet technology,social networks are quickly integrated into people's daily lives.The relationship between individuals is an important component of the social network platform to survive and develop.the rich information contained in the social network can provide strong support for public opinion monitoring,information dissemination research,advertising and so on.However,the rapid development of network technology has brought massive data and also has brought noise problems,such as false or the lack of information.How to use the observed data to predict or reduce missing information and mining implicit information becomes an important topic.Social network information mining is an important part of social network analysis,and it is also a hot research topic.In the real life,people connect with each other through different types of relationships.People and these relationships form the social networks that contain the same type nodes and various types of relationships between these nodes.This kind of network belongs to heterogeneous networks in complex network.At present,the research of social network analysis is mainly focused on the isomorphic networks.However,the research on heterogeneous network can find more accurate implicit knowledge.Therefore,this paper focuses on the relationship heterogeneous network.Due to family relations and social relations of its own characteristics,the different strategies used to mining and predict relationships in heterogeneous network.Overall,this paper contains the following three aspects:1?The Internet big data has large amount and many noise information,so we use crawler program to collect many famous people basic information and relationships from Baidu Baike.In order to store and utilize these data more efficiently,we use the graph database Neo4 j which can fully reflect the characters connections and relationship semantics.2?This paper analyzes the current situation of the research on family relationships reasoning,and points that most research surrounded with expert system or the Chinese language and literature.And that can not meet the needs of the application in the era of big data.In this paper,we used common sense and sociological knowledge to define the family relationships and its representation.At the same time,we use the first-order predicate logic form to formulate the basic rules of inference.The reasoning rules cannot be directly used in graph database operation,thus converting relation inference rules for database operation language.That can be easily to reasoning and complete family relationships.At the same time,three kinds of reasoning methods are adopted to meet the needs in different applications.3?In the aspect of social relations prediction,firstly,the research questions are described and defined.Secondly,analysis of the difference between this paper and link prediction in heterogeneous networks.Proposed a method to automatically find the path without setting the path mode,and to obtain the maximum reachable path from the importance of relationship paths.Thirdly,based on this method,a prediction algorithm for the social relationships in heterogeneous network is established,and realized the prediction of the social relationships.Finally,the proposed algorithms are compared with the traditional algorithm and further confirmed the validity and accuracy of this algorithm.