Design And Implementation Of Knowledge Fusion System For Domain Knowledge Graph

With the explosive development of the knowledge economy,knowledge graph plays an important role knowledge acquisition and display.In recent years,more and more research and implementation work has shifted from general knowledge graph to domain or industry knowledge graph.In order to improve the domain knowledge graph,professional data need to be input from different data sources.At this time,data redundancy or conflict may occur.In order to solve this problem,this paper designs and implements a knowledge fusion system for domain knowledge graph.This paper introduces the knowledge fusion system for domain knowledge graph in detail.According to requirements,this system is mainly divided into seven functional modules,which are semantic fusion management module,ontology model management module,knowledge graph data management module,multi-source data import module,knowledge graph sharing module,data statistics module and version management module.Through the basic functions of the seven modules,business personnel can input and integrate domain knowledge graph,and build a complete set of domain knowledge system.This system adopts Spring Boot micro-service architecture,and the backend is mainly implemented in Java.In terms of data storage,HugeGraph,an open source graph database of Baidu is used for data storage of graph data.MySQL database is used for data storage of semantic fusion data and system version.In terms of knowledge fusion,this system supports the import of multi-source data and provides four ways of knowledge fusion: semantic database matching,semantic similarity and attribute value combining to calculate weights,unique attribute value judging similarity,and using Node2 vec algorithm to vectorize nodes and computing cosine similarity.These four ways can eliminate semantic conflict and redundancy of entities and provide assistance for relevant business personnel to input and integrate domain knowledge graph.At present,the current system has been put into trial operation,and the system runs normally.After the actual test,the system performance is stable,and the input and fusion processing of the domain knowledge graph can be implemented.