Font Size: a A A

Research On Knowledge Graph Completion Method For Social Network

Posted on:2022-02-03Degree:MasterType:Thesis
Country:ChinaCandidate:J P XiangFull Text:PDF
GTID:2518306353977219Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
With the advancing of the times and the gradual development of the Internet,the data in social networks has become increasingly large,and the applications use social networks are also very extensive.However,most social networks are not complete,so it is of great significance to social network completion.Knowledge graph is an intuitive,efficient,accurate and suitable for processing multi-relational graphs of knowledge representation.The use of knowledge graphs to represent social networks and the use of knowledge graph completion methods to achieve social network completion have important research significance.The existing research on knowledge graph completion is mainly divided into rule-based knowledge graph completion methods and embedding-based knowledge graph completion methods.The former has the advantages of higher accuracy and strong interpretability,but poor generalization ability and high complexity.The advantage of the latter is better usability and scalability,but lower accuracy and poor interpretability.Aiming at the problem of social network completion,this paper combines two knowledge graph completion methods to propose a knowledge graph completion method for social networks.First,the social network space is constructed,the social network is transformed into a knowledge graph,and four relationship rules are defined: transitivity rule,antisymmetry rule,entity relevancy rule,and attribute relevancy rule.Secondly,use four relationship rules to propose a rule-based knowledge graph completion method,perform the first round of completion,and optimize the efficiency of the completion method.Thirdly,combining the rule-based knowledge graph completion method and the knowledge graph embedding method,the knowledge graph completion method based on embedding is proposed,and the second round of completion of the knowledge graph is performed.Improve the accuracy of knowledge graph completion,and optimize the efficiency of knowledge graph embedding training.Finally,a social network completion system was designed and implemented,providing visual displays,completion,analysis,editing and other functions on social networks,verifying the applicability and effectiveness of the proposed social network knowledge graph completion method.
Keywords/Search Tags:Social Networks, Knowledge Graph, Relationship Rules, Knowledge Graph Embedding
PDF Full Text Request
Related items