Research And Implementation About Semantic Analysis And Extraction Of Oil And Gas Exploration And Development Documents

The promotion and application of knowledge graph technology bring new research directions and challenges to related fields.The construction of knowledge graph in the field of Oil and Gas Exploration and Development(OGED)requires a large amount of knowledge,which comes mainly from related documents.However,there is no perfect method and system to automatically extract knowledge from these documents.So how to extract knowledge from unstructured documents on a large scale automatically has become the primary task for the research of knowledge graph in the field of OGED.In order to accurately extract the knowledge of OGED,this thesis studies the semantic analysis and extraction methods of OGED documents from three aspects:Information Extraction(IE),Named Entity Recognition(NER)and entity Relation Extraction(RE)of the documents.First,a method of IE based on rules and SVM is designed to extract the information of the documents,and the effect of the method is proved by comparison experiments.Second,depending on the characteristics of OGED,the appropriate rules and dictionaries are used to assist the machine learning to train the NER model,which further enhances the effect of NER in the field of OGED.Finally,A method synthesizing mode and distant supervision is used to extract entity relations,which effectively extracts a large number of relations in OGED domain.Through the integration of three parts of the study,the semantic analysis and extraction method of OGED documents has been proposed and realized.Experiments show that the method can extract a lot of professional knowledge automatically from OGED documents.