Construction And Application Of Knowledge Graph In Logging Domain Based On Deep Learning

Well logging is the process of recording different physical,chemical,electronic,or other properties of rock and fluid mixtures in a well drilled into the mantle.The logging data of field acquisition has achieved digital records,then produced a lot of well logging data,log data is an essential part of oilfield exploration and development valuable resources,but because of the lack of effective management of data,result in well logging data is too scattered,a lot of valuable data to query and use,there are a lot of well logging data is damaged because of improper management,The value of logging data is not fully exerted,which causes a lot of inconvenience to the daily logging related work.In order to solve the above problems,this paper constructed the knowledge graph of logging domain to manage the logging data,and realized the reliable storage and unified access of logging data,so that the logging data knowledge could play a better role.By referring to relevant literatures and research status of knowledge graph at home and abroad,theories and methods related to the construction of knowledge graph were deeply studied.In view of the current situation of logging data use and improper management,a knowledge graph of logging domain based on deep learning was constructed in this paper.The specific construction contents are as follows:Firstly,by studying the extraction method of logging entities,a deep learning model of logging entity extraction based on SENNA-BILSTM-CRF model is proposed,and logging entities are extracted from the data set.The extraction results verify the accuracy and effectiveness of the deep learning extraction model.Secondly,a deep learning model of logging entity relationship extraction based on Bert-Bigru-Attention model is proposed based on Bert-Logging-Bigru-Attention-FC by studying the extraction method of logging entity relationship.The triad set containing the logging entity relationship was extracted from the data set,and the triad set was added to the Neo4 j graph database by batch insertion to complete the construction of the logging domain knowledge graph.Finally,the logging domain knowledge graph system is developed based on the logging interpretation CIFLOC integrated platform,which takes the logging domain knowledge graph as data storage.The system provides visualization function of logging knowledge map,batch increment function of logging entity triad and maintenance function of logging knowledge map.