The purpose of text sentiment orientation analysis is to determine the attitude of the text by extraction and analysis. English text sentiment orientation analysis developed relatively early. The study of text sentiment orientation analysis of minority language is still in the initial stage of development. And the results of minority languageās sentence sentiment orientation analysis have not been seen. We proposed two methods to research Tibetan text orientation analysis.1. Tibetan sentence sentiment analysis that based on emotional dictionary.Firstly, emotional dictionary needs to be constructed in this method, and then the sentence sentiment by matching keywords is being calculated. During the experiment we constructed a Tibetan sentiment dictionary that has 5070 words. And the F value of this experiment is 68.9%.2. Tibetan sentence sentiment analysis that based onMaxEnt entropy.MaxEnt entropy is a method that based on statistical methods, and the experiment is based on MaxEnt entropy. During the experiment we built a corpus that has 10000 Tibetan sentiment sentences that had been tagged. The result of this experiment shows the F value of MaxEnt entropy is 82.8% which is greatly improved compared to the experiment based on emotional dictionary. |