Knowledge Graph Construction And Application For Enterprise Based On Multi-source And Heterogeneous Data

With the advent of the big data era,artificial intelligence technologies are driving the development of various industries toward digitization and intelligence.Knowledge graph is an important branch of AI,which can elaborate concepts and entities in the objective world and the relationships between them,and empower machines with cognitive intelligence in general or specific domains by building a huge knowledge network.At present,knowledge graph has been widely used in finance,justice,medical and other fields.Asset management Companies are mainly engaged in the management and disposal of non-performing assets,and it needs to conduct research on enterprises in the process of valuing assets.In the face of the huge amount of structured or unstructured data,enterprise knowledge graph provides an efficient way to manage and retrieve data,which can store the information from multiple sources into graph database after interaction and linking,and can visually present the correlations among enterprises and person.Based on an asset management company,this paper aims to build an enterprise knowledge graph by combining the company’s internal business data and the text data in the Internet.And on this basis,we build a system to help business personnel to refine enterprise related information more efficiently and carry out enterprise risk prediction and analysis of related enterprises.The main work of this paper is as follows.First,a large number of enterprise-related public opinion news and equity announcements are obtained through web news crawlers and Internet interface calls,and the data of specific several types of relationships are manually labeled.Based on these data,the CASREL model is implemented and improved as a joint entity relationship extraction model to solve the automatic extraction problem of relationship triples in unstructured financial text.Second,based on the internal structured enterprise business data and the information of enterprise entity relationship triad extracted from the unstructured text data,we design the fusion and update process of the data,construct the multi-source heterogeneous enterprise knowledge graph topdown,and store the graph data in Tigergraph graph database.Third,combining with the business requirements of company in the field of non-performing assets management,we conducted requirement analysis and outline design of the enterprise knowledge graph system,and developed the system using Springboot and Vue framework and completed testing.The system provides the functions of enterprise basic information query,news entity relationship extraction,enterprise relation graph query,enterprise guarantee chain graph query,etc.