Construction And Optimization Of Domain Knowledge Graph Based On Association Relationship Discovery

With the rapid development of artificial intelligence and big data technology,the number and types of information resources have increased rapidly.How to grasp the effective information in big data orderly and accurately has become a huge challenge.Knowledge graph can discover knowledge from scattered data,represent complex unstructured data into structured information,and help organizations realize business intelligence.Therefore,the construction and optimization of knowledge graph has become a research hotspot in many fields.However,the current research focuses on the extraction of simple entity relationship triples.There is a slight deficiency in making full use of domain-related data and constructing and optimizing domain knowledge graph by mining correlation between relevant data.The existing knowledge graph construction methods require a lot of manual annotation and expert knowledge,and mainly focus on the extraction of subject-verb-object relations.Therefore,it is difficult to effectively apply them to the construction of domain knowledge graph because they are often unable to mine hidden knowledge in complex data.To solve this problem,this paper proposes a knowledge graph construction method based on keyword cooccurrence algorithm and fuzzy string matching.Firstly,key words are extracted from the data related to the entity to obtain the key information related to the entity.Then,the relationship between the entities is obtained by analyzing the key information data.Based on the obtained key information,the fuzzy string matching algorithm is used to resolve the entity ambiguity.Experiments show that the proposed method can effectively mine the relationship between entities in complex data and accurately resolve the ambiguity of entities in the field.The existing knowledge graph construction methods do not fully consider the potential relationships between entities which hidden in large amounts of data,and the inferred relationships between entities have problems such as poor interpretability.To solve this problem,this paper proposes a method based on association analysis and topic analysis between entities,namely EA-LDA algorithm to mine the potential relationships between entities.Aiming at specific domain related data,the algorithm uses association rules to mine the association relationship between entities,and apply LDA topic extraction method to association rules to analyze the relationship between topics in the entity related data,and then obtains the hidden relationship between domain entities.The newly discovered relationships are integrated into the original domain knowledge graph through entity links to enrich the domain knowledge graph.Experimental results show that the proposed method can extend the relationship between domain entities in a more targeted and effective way,thus optimizing and perfecting the domain knowledge graph.