Application of PMI Algorithm in Text Mining

摘要

In the Chinese opinion mining, the relevant scholars will focus on how to accurately receive the semantic emotional of opinion word as their breakthrough points currently, but the accurate access to features and characteristics of the relationship between the relatively was few studied. Correlation level analysis of characteristics will play an important role in the following semantic emotion analysis and understanding of the entire review. Based on the hierarchical algorithm of feature extraction in the past, this paper proposed an improved method. After the analysis of specific grammatical structure in Chinese, the algorithm finds whether there are different characteristics of hierarchical relationships between the words with specific grammatical structures and Chinese Internet commercial searching engine results. Analysis shows that network-based PMI algorithm can effectively obtain the characteristics of hierarchical relationships between the words.