Research And Implementation Of Knowledge Extraction In Exploration And Development Documents

Over the years,the introduction of knowledge maps has made search from linking texts to linking data,pointing directly at the results of the answer itself,making it easier for computers to analyze and process data and establish relationships between entities.However,the current knowledge maps in the field of exploration and development are still not perfect.There are not enough entities and relationships in the knowledge maps.Therefore,how to extract the relationship between entities from the exploration and development documents as much as possible has become the primary task of developing a domain knowledge map.This article mainly analyzes complex sentences that have multiple semantics and no punctuation.In order to extract the knowledge in the sentences as completely as possible,this paper is divided into three steps.Firstly,we proposed a method of acquiring corpus in combination with rules and manual collection,and constructed a Chinese syntactic dependency tree to obtain training sets and train action classifiers,and then obtain a training set and train action classifiers;secondly,we proposed a method to classify edges in the Chinese syntactic tree by using an action classifier to obtain clauses,and successfully transform complex sentences into multiple simple sentences;finally,we use the methods based on syntactic dependency,remote supervision and LTP to extract the entity relations of the obtained simple sentences and original sentences and analyze the results.Experiments show that the method of extracting corpus from complex sentences into simple sentences and then extracting entity relations can improve the integrity of extracted knowledge more than directly extracting from sentences.