| With the rapid development of the Internet,more and more ethnic minority people use micro-blog to express their opinions,express their feelings and exchange their experiences.The study of Tibetan texts can not only understand the daily life and viewpoint expression of ethnic minorities,but also obtain the attitudes of minority users towards certain types of services and products.It will also be beneficial to the improvement of the quality of social services in minority areas and the promulgation of reasonable policies.Based on the construction of Chinese-Tibetan sentiment dictionary,this paper constructs the emotional computing model of Tibetan micro-blog text.The research content mainly includes:(1)It summarizes and analyzes the current emotional dictionary construction and the techniques of emotional calculation of foreign,Chinese and Tibetan texts and the latest research results.(2)Using the existing Chinese sentiment dictionary resources to automatically construct the Chinese-Tibetan bilingual sentiment dictionary not only solves the problem of time-consuming and laborious construction of Tibetan dictionary,but also ensures the correctness of Tibetan sentiment dictionary and enough vocabulary.Based on the constructed sentiment dictionary,the emotional calculation of Tibetan micro-blog text is carried out.The experimental result has an accuracy rate of 76.6%,indicating that the sentiment dictionary has reached practicality.(3)Aiming at the small problem of the original data set,the Tibetan and antonyms dictionary was constructed and the data was automatically expanded by the same method of replacing the antonym.(4)According to the characteristics of convolutional neural network and long-term and short-term memory network,the CLSTM hybrid deep learning model is proposed,and a good emotional calculation effect is obtained.The accuracy rate reaches 85.1%.The problems of model parameter optimization and so on are studied to achieve the optimal effect. |
PDF Full Text Request