Research on Tibetan Text Orientation Identification

In recent years, Minority languages in China are widely used on the computer and network. But now there is no effective public opinion analysis system of the minorities overall attitude of the masses of the hot events or topics. In this study, we research on Tibetan topic orientation recognition. First, according to the Tibetan context and life characteristics, combined with a set of emotional words in Hownet, the Tibetan emotional word dictionary is built, and then by the Tibetan word semantic similarity calculation method we extend this dictionary to get rich emotional word set. We also propose a method that the sentence orientation is determined by the orientation of words in this sentence and the orientation of text is determined by the orientation of sentences in this text. By our research the Tibetan hotspot information can be rapidly detected and found and then the public opinion tend can be track quickly. It is benefit for positive guidance of public opinion.