Research And Application Of Domain Oriented Entities And Inter-entity Relations

In recent years,the rapid progress of artificial intelligence technology has promoted the development of all walks of life,leading to continuous breakthroughs in technological applications and accelerated updates of technological products.At the same time,with the vigorous development of the Internet and big data,the development of the industry has gradually become transparent and popular,and more and more people interested in related knowledge have begun to pay close attention to the development of technology and the birth of new products.Nowadays,in the face of the increasing speed and number of professional thesis published,how readers can find useful information from massive thesis and how to comprehensively understand highly professional thesis have become hot issues that need to be solved urgently.This thesis analyzes the current research status of named entity recognition,relationship extraction and reading help system at home and abroad.In the entity relationship extraction of deep learning,entities and entity attributes are mostly displayed in the form of nouns,and there are structured information of abstract-concrete relationship and whole-part relationship in nouns.The thesis finds an organic combination point from deep learning relationship extraction and professional structured information,namely abstract-concrete relationship and whole-part relationship,and applies it to text information extraction.Based on this,it establishes a domain-based knowledge Deep learning text comprehension help system.In the implementation of the system,data sets related to the field were collected and produced,and the labeling of the data sets were analyzed,and the labeling methods of professional entities were changed,so that the recognition effect of the deep learning model on professional entities was improved.A deep learning model based on BERT is constructed,and the multi-label classification based on BERT is used in the model to classify the entity relationships in the text.Moreover,this model uses K-BERT,which introduces knowledge graphs,to perform named entity recognition on the text.Aiming at the characteristics of human cognitive process and professional knowledge,this article explores the abstract-concrete relationship and the whole-part relationship between professional entities and entities,and applies abstract-concrete knowledge and whole-part knowledge in the help system to build Abstract-specific knowledge base and whole-part knowledge base to help interpret the professional knowledge in the article.And the division level of abstract-specific knowledge corresponds to the user's professional knowledge reserve.In the interaction with the user,the user's understanding of the professional knowledge,that is,the user's professional knowledge level,is obtained.Combine the knowledge base with the deep learning model,and provide personalized help for different users according to their professional knowledge level,so that readers can further understand professional knowledge on the basis of their own knowledge reserves.At the same time,in order to give users a good user experience,the system implements an interactive interface,and designs an input module,an output module and an interface for obtaining user professional knowledge levels.Among them,the design of the output interface uses a force-directed graph to display the results.Finally,in the field of instrumentation,the prototype system of the deep learning text understanding help system based on domain knowledge is realized.Through experiments,the information entered by the user is obtained from the interactive interface,the entity relationship is extracted from the article uploaded by the user,and the technical products and professional entities in the article are supplemented with personalized knowledge from a professional perspective,and then the knowledge is displayed to the user and it verifies the feasibility of the model.