Research And Application Of Knowledge Graph Technology For Big Data Of Credit Investigation

Big data of credit investigation is characterized by complexity and high dimension,and there are various complex relationships and concepts among credit investigation entities in the real world.Knowledge mapping can reasonably represent entities in the real world and their associated relationships,and accurately display semantic relationships between concepts.The construction of credit investigation knowledge atlas can extract all kinds of knowledge implied in the big data of credit investigation and the association between knowledge,and mine more potential information resources.At present,the knowledge atlas in the field of credit investigation generally has some problems and challenges,such as low degree of specialization,unclear data hierarchy,difficulty in cross-domain joint analysis,and insufficient knowledge value mining.The credit investigation knowledge atlas is mainly divided into individual credit investigation knowledge atlas and enterprise credit investigation knowledge atlas.This paper focuses on the construction of enterprise credit investigation knowledge atlas based on credit investigation big data.Specifically,the difficulties in constructing the knowledge map of enterprise credit big data are as follows:1)In view of the strong privacy of enterprise credit big data and its typical characteristics of mass,multi-source and heterogeneity,there is a lack of unified organizational standards and specifications for enterprise credit big data resources as theoretical guidance for the construction of enterprise credit big data knowledge atlas.2)Due to the lack of theoretical guidance and data support,there is currently a lack of ontology and examples of credit investigation knowledge atlas constructed based on enterprise credit investigation big data.In the risk control model based on the enterprise credit atlas,although the current main line method extracts certain enterprise relationship features based on the atlas structure,due to the incomplete construction of the atlas and traditional cognitive problems,the enterprise entity attribute features extracted from the atlas are often focused on the financial and judicial fields.Due to the few data dimensions selected,the prediction effect of the model needs to be improved.3)There are complex correlation relations between credit investigation entities,and it is difficult for people to clearly define the correlation between credit investigation entities and dig more potential information through traditional representation methods.In view of the above problems and challenges,this paper focuses on the unified information model of enterprise credit investigation big data,the construction and risk control model design of enterprise credit investigation big data knowledge graph,and the design and implementation of enterprise credit investigation big data knowledge graph platform.The main contents are divided into the following three aspects:1)Propose and implement a unified information model of big data for enterprise credit investigationFrom existing government affairs,industry and commerce,justice,public opinion and so on various sub domains dispersed data extract"enterprise-a key figure" joint framework,design a set of enterprise credit data scenarios hierarchical enterprise information architecture and framework of unified information model,key personnel to the current enterprise massive,multi-source and heterogeneous data in the field of inquiry,And provides a basis for the establishment of standards and norms in the field of enterprise credit investigation.2)The knowledge graph of enterprise credit big data is constructed and a risk control model based on the graph is proposedThe construction method of the ontology and data of the knowledge atlas in the credit investigation field,which is dominated by top-down and supplemented by bottom-up,is adopted to complete the defects of the single construction method and help form the knowledge atlas of big data in the enterprise credit investigation with clear organizational structure and strict hierarchical constraints.Is put forward based on the enterprise credit risk control model of big data knowledge map,in the mainstream enterprise basic attributes and knowledge map network characteristics(the relationship between the enterprise and the rest of the entity),based on the innovative introduction can reflect the characteristics of enterprise innovation ability,such as r&d,patent number as the characteristics of the model,achieve the risk assessment of the enterprise,The average accuracy and AUC of the model reached 0.85 and 0.93.3)Build the enterprise credit big data knowledge atlas platformEnterprise credit reporting big data knowledge map platform using data visualization technology to clear the basic information of the present enterprise and the relationship between the enterprise and the rest of the entity,present a visual image of complex information,make the practitioners clear understanding of the enterprise information and relationship,help them to enterprise risk prediction,association analysis and application.