Knowledge Graph Construction Method And Application In The Field Of Oil And Gas Exploration And Development

It has been more than 40 years since information construction was carried out in the oil and gas exploration and production.Large number of information models have been established to support various scientific applications.A mass of “information islands” emerged because of lacking of uniform standards between different disciplines during the establishment of information models.The emergence of "information island" makes it difficult to share information for the disciplines and specialties.In this case,knowledge base is badly in need of supporting oil and gas exploration and production,which is coupled with the application of artificial intelligence and big data.So,knowledge model will be constructed where massive,dispersed and heterogeneous information of exploration and production are organized.However,the standardization of knowledge representation in oil and gas exploration and production has not been established.In addition,it is difficult to implement semantic retrieval and knowledge recommendation of data based on oil and gas exploration and production information model.A method of standardized domain knowledge graph construction based on metadata registries(MDR)is proposed to solve the problem of standardized representation of knowledge,where the knowledge graph technology and ISO/IEC11179 data element registration standard are studied.On this basis,the query expansion method and the personalized recommendation application of domain semantic data are researched.Main contributions of this thesis are as follows:Firstly,a new method of standardized domain knowledge graph construction is proposed on basis of metadata registries(MDR),where knowledge graph technology is used to solve the problem of standardized knowledge representation.Two standardized mapping rules,MDR conceptual model and concept relationship graph,data description metamodel and concept relationship graph,are established separately,where framework model of MDR generating knowledge graph is set up.And the conceptual relationship graph of knowledge graph is constructed.Mapping rules and mapping algorithms are proposed between relational database and entity relation graph,which are used to release the entity relationship graph based on relational database.Standardized construction of domain knowledge graph has been realized,which solves the problem that the construction methods of domain knowledge representation method are not uniform and standard,and that semantic knowledge cannot be shared and reused in the domain.Secondly,a semantic query expansion algorithm,based on the integration of comprehensive similarity calculation and knowledge graph technology,is proposed in the frame of standardized knowledge representation where several conceptual similarity algorithms are compared and analyzed.The new algorithm can be utilized to solve the problem that the retrieval content of users cannot be expanded effectively and the query results are one-sided,and realizes the query expansion of domain semantic data.Semantic data query experiment is carried out based on the actual conceptual relationship weight and business requirements in exploration and production domain.The experimental results show that the accuracy and recall of user data query under appropriate thresholds will be improved.Thirdly,several personalized recommendation algorithms are analyzed comprehensively,and a hybrid recommendation algorithm based on knowledge graph and collaborative filtering recommendation algorithm is proposed.Combination of knowledge graph and K-means clustering algorithm is introduced to choose similar users,and figure out the problem of sparse data of user-interest matrix.And user-based collaborative filtering recommendation is realized by SVD decomposition method,which solves the problem of personalized information recommendation.The personalized information recommendation experiment is carried out by combining the actual business needs and data in the exploration and production domain.The experimental results demonstrate that the accuracy of information recommendation is enhanced.Finally,the application of knowledge retrieval of formation data is designed and developed based on the requirement of geological design for underground operation of geological brigade for the retrieval of formation knowledge.Based on MDR in exploration and production domain,the conceptual relationship graph of formation semantics is constructed.The data related to formation in EPDM are mapped to entity relationship graph.By using the query expansion and personalized recommendation algorithm proposed in this thesis,the intelligent retrieval and push of the semantic knowledge of formation data are realized.The practical application proves the feasibility of the knowledge graph construction method and application algorithm in this thesis.At the same time,the knowledge graph construction method and its application in query expansion and personalized recommendation are universal,whichcan provide reference for the construction and application of knowledge graph in other domains.