Tibetan Micro-Blog Sentiment Analysis Based on Mixed Deep Learning

The analysis of micro-blog sentiment tendency can reveal the changes of people's emotions, which has become a hot topic in text research. There is a relative lack of research on Tibetan language. In this paper, deep learning algorithm is introduced into the Tibetan sentiment analysis to study the accuracy of different algorithms on Tibetan micro-blog sentiment classification. Firstly, the Tibetan word micro-blog is trained as a word vector by using the word vector tool; then the trained word vectors and the corresponding sentiment orientation labels are directly introduced into the different deep learning models to classify the Tibetan micro-blogs. The results show that the hybrid deep learning algorithm obtains a good classification effect. Different optimization parameters were studied on this model, and the accuracy was improved by 1.22% compared with the previous one.