A Study On The Extraction Of Basic Concepts In Domain Sentences

Text as one of the important media people use language communication plays a vital role in people's historical activities.Text representation as the first step in the machine's understanding of the text is also becoming more and more important.The current semantic representation is mainly based on words or sentences as the basic unit.This approach is contrary to the way people understand the language,and the effect is poor.Therefore,the way of simulating the natural language is simulated and the key information in the sentence is extracted.The basic concept of the structure of the sentence,and try to use this concept structure to carry out semantic representation,is a potential research approach.In terms of research methods,some of the existing natural language processing methods are gradually evolving from the previous rule-based approach to the current statistical-based approach.In the future,how to use the massive data provided by the Internet and use the computer in the large data background Intelligent algorithm to achieve the text of the representation and semantic excavation,has gradually become the focus of research and direction.This paper first studies some basic steps in text processing,introduces some past text representation methods and commonly used algorithm models in semantic mining.For some common classification models,including Naive Bayesian algorithm,support vector machine,K neighborhood classification Algorithms,as well as deep learning areas are very popular in the depth of the belief network model,also carried out research.Then,based on the theory of conceptual map in text representation,this paper proposes a Chinese semantic chunk for Chinese text representation.It aims to combine the Chinese semantic chunks to complete the conceptual structure of text sentences.In this paper,we construct a semi-automatic corpus generation model with support vector machine as the core algorithm,and construct the relevant corpus of Chinese semantic chunks in the field of science and technology review.At last,we study the deep learning model represented by deep belief network In this paper,the corpus constructed in this paper is studied,and the feasibility of the deep learning algorithm in Chinese semantic chronological extraction is studied.The corresponding model is constructed.By comparing the results of Chinese semantic chunks under different algorithms,the depth learning algorithm In the field of text extraction good results for the natural language to deal with the text to provide a new way of thinking.