Research On The Method Of Extracting Enterprise Tax Law Entity Relationship Based On Dependency Parsing

With the continuous development of artificial intelligence technology,the construction of man-machine dialogue system has become a hot spot in the field of research.The realization of man-machine dialogue can not be separated from the processing and understanding of text information(hereinafter referred to as “text”)by computer.According to the existing research,the understanding of text by computer is not intelligent enough.In order to achieve intelligent text processing,computers need to analyze both syntax and semantics.When confronted with ambiguity,they also need to deal with the context.This paper is based on the above principles.This paper focuses on named entity recognition and entity relationship extraction.In order to achieve better results,it is necessary to combine grammatical rules with semantic analysis.Named entity recognition is summarized and sorted out by knowledge map,which is displayed in a visual form.At the same time,it ensures the comprehensiveness of named entity to the greatest extent and finds the missing content in time.To extract entity relations,we need to start with grammar and semantics.In order to obtain better grammar rules,this paper chooses dependency syntax to analyze them.This method can be used to deal with long and difficult sentences and ambiguous sentences,and can better show the core and hierarchy of sentences.At the same time,the extraction of entity relations can not be separated from the semantic mapping of entity relations words,which requires the semantic classification of entity relations,rather than just staying at the grammatical level.With the rule analysis at the grammatical level and the classification of entity relations at the semantic level,we can better understand the true meaning of a sentence,and thus realize the processing and recognition of text information.The research area of this paper is enterprise tax law,so the research goal is related entities and entity relations involved in the field of company tax law.This paper refers to six professional books in this field,and through Baidu entry,Baidu question and answer,Zhihu platform to search for relevant tax laws related to enterprises,summarizes,classifies and categorizes the named entities involved.In the construction of entity relationship system,a corresponding vocabulary is also established for each entity relationship.Mind Manager(Mind Map)-screen is used as a tool to generate entities and entity relationships involved in this field,and then they are displayed graphically.At the same time,we extract the named entity relationship involved in it.The extraction mainly summarizes the syntactic rules,parses the new sentences and gets the corresponding participle.After word segmentation,part-of-speech tagging and parsing,each sentence will have a corresponding syntax tree,and each syntax tree can be represented by corresponding tags.This paper chooses the field of enterprise tax law as the research object,which is inseparable from its importance.Taxation comes from the people and serves them.Taxation plays an important role in the development of national or government revenue and social economy.The basis of taxation is the tax law.Based on this,the author believes that the study of enterprise tax law has very important social value.At the same time,at this stage,the whole society has entered a state of nationwide company-opening.For entrepreneurs,they urgently need to learn the relevant knowledge of corporate tax law,and the research on "corporate tax law" in this paper will provide them with a platform to understand the knowledge in this field.In addition,based on the research of extracting the entity relationship of enterprise tax law and the current research hotspots,this paper combines the research content with the automatic question answering system,so that the research results can be applied to practice,and also makes the research have more practical value.To sum up,this paper applies dependency parsing technology and semantic classification to the field of enterprise tax law,trying to achieve the goal of automatic extraction of named entities and entity relationships.On this basis,a set of ideas for building an automatic question answering system in this field is put forward,and it is hoped that with the help of this research,it can be applied to the actual automatic question answering system.