Tibetan sentence sentiment computing based on statistics

摘要

This electronic document is a “live” template and already defines the components of your paper [title, text, heads, etc.] in its style sheet. Sentiment classification is one of the research hot spots of Natural Language Processing. Compared with English and Chinese, it is hard for Tibetan to do some research of sentiment analysis because of the situation that we are lack of related sentiment corpus. In this paper, we construct a Tibetan sentiment corpus by crawling from Tibetan website and artificial Chinese-Tibetan translation. The final corpus we built is basically reaching an experimental requirement. The corpus contains 10,134 Emotion sentences, including 2,025 artificial translation corpuses, and 8109 corpus crawling through the network. The Tibetan sentiment sentences are computed by learning from the Chinese and English sentiment analysis based on statistical method. We established the Tibetan sentence sentiment analysis system based on the maximum entropy model, and the system is test by use of the corpus we built. Results show that the evaluation parameter is good and it can achieve practicability to some extent.