Domain-specific Expert Knowledge Graph Construction And Disambiguation

With the development of Internet technology and the advent of the era of big data,expert information in various fields can be easily found on the Internet.However,due to the uneven quality of data sources,the description of different data sources about the attribute value of the same expert may conflicts.When selecting experts and tracking research teams,it is necessary to use complete and accurate expert resumes.However,there are few studies on the construction of expert resumes and disambiguation work.The inaccuracy and dispersion of information make it difficult for users to obtain accurate and complete expert information.The knowledge graph can construct entities,attributes,and attribute values together,and can organize expert resume information well.Therefore,this paper uses knowledge graphs to construct expert resumes,gathers dispersed information into the same knowledge graph,and proposes a search engine-oriented triple scoring model to score conflicting data.After improving the model,this paper proposes a domain-based scoring model.In addition,this paper proposes a truth verification algorithm combined with the scoring model,through which accurate and complete expert resumes can be obtained.The main contributions of this paper are as follows:(1)Aiming at the problem of information dispersion,this paper proposes a method for automatically constructing a knowledge graph.This method can extract the entity,relationship,attribute triples of experts from different encyclopedia websites,and combine the triples into a resume,then merge resumes from different data sources to form an expert knowledge graph.(2)Aiming at the problem of attribute value conflict,this paper designs a search engine-oriented scoring model,which uses the search results of the search engine as the data source to vote for the attribute values in the true value candidate set.Since this model is a universal model in various fields,and the credibility of the specific data source of the expert's field should be higher than that of the general data source,this article improved the model and added the influence factor of the field.Experiments show that the accuracy of the attribute values processed by this model is higher than that of the original data source.(3)Most of the existing truth verification algorithms are aimed at the disambiguation of single truth or multiple truth,and this paper proposes a universal truth verification algorithm.Just input the constructed expert knowledge graph,it can automatically extract the conflicting attribute values from the resume,and use the voting model to score the conflict data,and finally output the expert resume with the candidate value score.(4)This article implements a field-oriented expert knowledge graph disambiguation system,using this system,users can automatically generate the expert knowledge graph by inputting an expert name.Construct and disambiguate expert resumes can provide users with objective and accurate data so that users can make the right choices.The system functions mainly include viewing existing expert resumes,viewing expert resume information circles,generating and disambiguating expert resumes,viewing expert resumes by field,etc.