Construction Of Web Community Text Entity Relations Map Based On Semantic Elements

It is a hot and difficult point to analyze the content of the text contained in the network community.On the one hand,the contents of the network community is different from the standard news and medical texts.Its characteristics of colloquialism,short text,noise and non-standard text make it difficult to mine public opinion information.On the other hand,most of the mainstream text information extraction and mining technologies are based on the shallow semantic level of keyword matching and topic clustering.Its performance is unsatisfactory when identifying the public opinion information contained in the network community text,and it is difficult to detect and warn the public opinion timely and accurately.In order to improve the accuracy of public opinion recognition in the field of network community,this paper first studies the text characteristics of network community,and considers how to deeply understand the text content of network community.On the basis of literature research at home and abroad,the research direction determined this research topic,that is,by mining the deep semantic relationship contained in the network community text and constructing a visual atlas to characterize and identify the public opinion information in the network community text.This paper designs a technical framework and implementation process which includes three parts:domain core semantic model construction?network community text analysis and semantic model fusion and presentation.Firstly,aiming at the typical network community college public opinion field,this paper uses semantic atlas technology to build core deep semantic model;secondly,it uses improved entity recognition and entity relationship extraction technology to analyze the contents of university community which contains public opinion and get the entity relationship contained in the text.Finally,to integrate the entity relationship and the core semantic model of the network community university which contains public opinion,and construct a deep semantic relationship and dynamic expandable entity relationship map which can represent the network community university public opinion field and then visualize it.In the implementation process of the specific technology,this paper verifies the accuracy of the improved entity recognition and entity relationship extraction technology by designing comparative experiments,and designs corresponding verification methods to verify the accuracy of the final visual atlas.Experiments show that the network community text entity relationship map based on semantic elements can better represent and recognize public opinion information in specific areas of the network community.