The research of estimation model for the correlativity between words in Chinese text

The analysis and use of the relation between words in Chinese text by statistical method is discussed in this article. After understanding the importance of the relation between words we investigated the characteristics of measurement models such as mutual information and related degree, and constructed a cascade estimation model which is used to describe the bi-orderly-neighbor between neighboring words in Chinese. Then, based on the characteristic of the Chinese text proof distance information between words and context information of current word were combined with N-gram model, and a novel model of correlativity between words based on language environment is present. Finally, the two models were applied in Chinese text automatic proof and corresponding experiment data and results are presented. Results show that the language environment related model can describe the relation between words better than other models such as mutual information or related degree, and indicate a good effect in the automatic defection of Chinese text errors.