Research And Implementation Of Enterprise Profile Technology Based On Knowledge Graph

Research into enterprise profile has begun to appear in recent years,and products related to enterprise profile have been applied to government bodies and enterprises.However,many such researches just show the enterprise information in map or chart by connecting to specific enterprise information data source.In fact,enterprise information includes a large amount of hidden knowledge.Based on enterprise knowledge graph,the paper integrates enterprise information of many aspects to portray an enterprise by inferring knowledge hidden under the surface of the enterprise information.The paper puts forward enterprise profile based on knowledge graph.Enterprise knowledge database is established by integrating knowledge graph with enterprise information obtained from many data sources on the internet.Also,knowledge is learnt by using Markov logic network to supplement the knowledge.Generally,the enterprise knowledge graph established in the paper can be divided into three levels: the relation of enterprise entities at the bottom level,the entity of Markov logic network at the middle level and the knowledge of Markov logic network at the upper level.The clause can be used in reasoning of Markov network to supplement the knowledge graph,and can also be used to dig knowledge hidden under the surface of enterprise information,such as research orientation of the enterprise and abnormal enterprise employees.The enterprise profile is drawn based on the enterprise knowledge graph established in the paper.The paper concludes the pattern drawing and designs predict based on available data of the enterprise knowledge database,and uses knowledge learning algorithm Markov logic network,whose name is DSL algorithm,thus obtaining the carrier clause of the knowledge and the weight.The entity relation of basic enterprise knowledge database is sparse without a clear layer,and some of th e entity relation can be supplemented by knowledge reasoning,using Markov logic network.The paper realizes the maximum problem reasoning algorithm LazySAT of Markov logic network to infer relation among entities,with clauses learnt from Markov logic network their weight as the input.The key of enterprise profile lies in the direct display of enterprise information,including basic enterprise information and inferred enterprise knowledge.Taking an enterprise as an example,the paper analyzes from three perspectives of enterprise scientific research orientation,remark of employees to the enterprise and employees frequently change jobs.The technology of enterprise profile can be applied to business scenarios such as investigation of enterprise background,search of enterprise partners and discovery of enterprise problems.