Research And System Implementation Of Chinese Lexical Substitution

The goal of lexical substitution task is to replace the target word in the sentence with substitute words without changing the meaning of the sentence in a given context.The research on lexical substitution abroad has been carried out earlier,and more research results have been achieved.lexical substitution has always been a key research direction in the field of natural language processing,but because of the complexity of the problem,the problem of Chinese lexical substitution has not been well solved.With the continuous introduction of algorithm theory and technology,it is gradually possible to solve the problem of Chinese lexical substitution.The results of this study can be applied to many fields of natural language processing and generation,such as lexical simplification,word sense disambiguation,abstract generation and so on.Focusing on the task of lexical substitution in Chinese composition,this paper makes a lot of exploration on the theoretical and practical problems involved.Firstly,it constructs the Chinese composition lexical substitution model based on dictionary as the baseline model,and then constructs the Chinese composition lexical substitution model based on neural network.On the basis of the research results,A Chinese composition lexical substitution assistant system is developed.Its main goal is to identify and replace the words that are not suitable for the context or are not wonderful enough in the text,so as to improve the quality of composition.In this paper,the main contents and innovations of this paper are as follows:（1）In order to solve the problem of lack of Chinese lexical substitution research resources,a data set of Chinese lexical substitution is constructed.In this paper,intermediate and primary vocabulary in HSK and the Modern Chinese Language Corpus and Chinese translation corpus of The State Language Commission are used as the data sources for constructing the dataset.Referring to the construction of English lexical substitution data set,following the construction steps of putting forward sentences,providing alternative words and merging notes,the Chinese composition lexical substitution data set chls is created,The data set contains about2000 pieces of data,which can be well applied to the research of Chinese lexical substitution.（2）In order to solve the problem of lack of research on Chinese lexical substitution and further study Chinese lexical substitution,this paper draws lessons from the methods used in English lexical substitution task,proposes and constructs a Chinese lexical substitution model based on dictionary as the baseline model of Chinese lexical substitution research.In order to improve the efficiency of Chinese lexical substitution,This paper also constructs a Chinese lexical substitution model based on context aware word embedding,a Chinese lexical substitution model based on Bert language model and a Chinese lexical substitution model based on fusion to improve the replacement efficiency of Chinese lexical substitution.The experimental results on the Chinese composition lexical substitution data set chls show that,The model proposed in this paper has achieved good results in Chinese lexical substitution.（3）In order to make the research of Chinese composition lexical substitution well applied,this paper combines the lexical substitution model based on dictionary and the lexical substitution model based on neural network,designs and implements a Chinese composition auxiliary lexical substitution system using flask framework and B / S architecture.The system will automatically replace the words that are not suitable for the context and wonderful in the composition,and generate a personalized report for users.