Research And Implementation Of Entity Relation Extraction Method In Oil And Gas Exploration Field

Entity relation extraction is one of the core tasks of information extraction.Its goal is to identify entities from the text and extract the semantic relations between the entities.Entity relation extraction is an important step in building a knowledge base and knowledge graph,and is widely used in intelligent question answering,relationship-based search engines,and machine translation.However,compared with the general field,the extraction of entity relation in the field of oil and gas exploration faces the challenge of lack of training data.The manual labeling of training data is expensive and the existing labeling methods cannot automatically label large amounts of high-quality training data.On the other hand,there are many types of entities and relations in the oil and gas exploration field,and it is difficult to predefine all the types.In view of the actual situation in the field of oil and gas exploration and the deficiencies of existing research,this paper mainly carried out the following two aspects of research:(1)In order to solve the problem of complex entities and relations in the field of oil and gas exploration and lack of large-scale training corpus,this paper proposes an open entity relation joint extraction model CSSEM(Chinese Syntactic Structure Extraction Model)based on syntax structure.CSSEM uses syntactic information and automatically learns entity and relation extraction patterns based on a small amount of labeled corpus.On this basis,an entity recognition method based on extraction patterns is given.Finally,combining the entity recognition method and relation extraction patterns directly extract entities and relations from the unstructured text.Based on evaluation criteria such as accuracy,recall,F-value,AUC,etc.,CSSEM and other models have been experimentally compared on datasets in the fields of news,biology,and exploration.The results show that CSSEM has achieved good results.(2)In order to use the auxiliary information in the text to further improve the quality of entity relation extraction,we proposed a Chinese extraction model CREBAI(Chinese Relation Extraction Based on Auxiliary Information)using auxiliary information,which uses convolution neural network encodes sentences from text and uses additional auxiliary information mined from the training data to improve relation extraction.In addition,on the basis of the CSSEM model extraction patterns,we propose an algorithm ATDBOPs(Annotating Training Data Based on Patterns)for automatically labeling data based on patterns,which provides training corpus for CREBAI model.CREBAI compared with other five models on two benchmark datasets.The experimental results verified that CREBAI used auxiliary information to further improve the performance of relation extraction in the field of oil and gas exploration.