Research And Implementation On The Construction Method Of Domain-oriented Knowledge Base

With the advent of the Industrial Internet,countries around the world are carrying out a digital revolution around industrial manufacturing,accelerating the construction of a new industrial ecology featuring intelligent production,personalized customization,and collaborative production.The domain knowledge base is a special and commonly used database,and efficient construction of domain knowledge base is the main difficulty of current research.However,in previous studies,most scholars have ignored the detailed content that the ontology can express.The domain knowledge is complex and diverse,and the existing ontology description methods cannot display domain knowledge well.In addition,the Industrial Internet is a brandnew industrial ecology.Industrial mechanisms are very common in industrial production.They contain logical rules,formula theorems and other forms.These features are difficult to describe in the ontology,and industrial mechanisms are dynamically changing.Therefore,this article will analyze from the perspective of ontology modeling and study the construction method of domain-oriented knowledge base.In order to research the method of domain knowledge base construction,this thesis firstly proposes a three-layer model suitable for domain ontology modeling.Then,the operation of the three-layer model is studied,and the operation set of the three-layer model is proposed.Secondly,based on the three-layer model,research on the construction method of the domain knowledge base,focusing on the problem of data extraction at the capability layer,proposes a construction method of the domain knowledge base based on the three-layer model,and takes the clothing domain as an example to verify.Finally,a knowledge base construction system is designed and implemented.The specific research content is as follows:For the modeling method of ontology,this article firstly defines the metametamodel and describes the meta-metaobject.Then the meta-model is defined based on the meta-metamodel.Secondly,it defines the ontology modeling framework of the three-layer model.The three-layer model is hierarchical and multi-view.According to the applicable scenarios,the composition view,capability view and mode view are defined.Finally,the basic operations and view operations of the three-tier model are proposed.For the construction method of the domain knowledge base,this thesis proposes a construction method of domain knowledge base based on the three-layer model,and takes the clothing domain as an example to verify it,and introduces the construction process of domain knowledge base in detail from the processes of data processing,ontology construction,and knowledge extraction.Then based on the operation set of the three-tier model,six applications are introduced,and the operations are verified in the constructed knowledge base.For the problem of content extraction in domain data,this thesis implements a relation extraction model based on CNN based on the open source project DeepKE.Taking the clothing field as an example,the data set selects clothing field data and manually marks the data.By using sampling probability to alleviate the problem of uneven category,complete the training of the model.The accuracy of the model is95.71%Based on the theoretical research of the above-mentioned domain knowledge base construction method,a knowledge base management system is designed and implemented.Mainly realize domain ontology construction,domain knowledge extraction and domain knowledge base construction.For domain ontology building blocks,use Web VOWL for visual display.For the domain knowledge extraction module,the knowledge extraction model of the encyclopedia and the knowledge extraction model of the clothing field are mainly used,and Docker is used for containerized deployment.Through the knowledge extraction model,the head entity,tail entity and relationship can be extracted,and JSON files can be exported,developed into middleware,and published as RESTful services.For the domain knowledge base building module,the graph database Neo4 j supports importing JSON files,which improves the construction efficiency of the knowledge base.Finally,functional testing and non-functional testing of the entire system have verified the feasibility of the above method.