Research Of Emotion Recognition System Based On Chinese Text

With the development of computer network communication and multimedia technology, Human Machine Interaction(HCI) technology has become a quite active research in the computer field. The research of affective computing can enhance the computer's intelligence and humanity to develop new man-machine environment to promote the development of psychology, intelligence and other subjects. It hasimportant practical significance and great economic and social benefits.Using the computer to simulate the human psychological behavior has been pursued by the people. At present, the relevant technical studies are forming a trend. Japan, the United States and other developed countries have begun to do emotional field of science research and exploration. The purpose of affective computing research is making machines identify, understand, express and adapt to people's emotion. If the computer is given ability of feelings similar to human, sense and expression of emotion, man, nature and machine would reach a harmony of the interaction.First, the research background of affective computing, the key technology of affective computing and the relevant affective computing are introduced in this paper. And the technology of emotion recognition, modeling and syntax analysis are elaborated in detail. Besides, the research and development of this field in domestic and foreign are related. Second, the methods of analysis, the design for emotion recognition of Chinese text are studied. We pay more attention to the process of analysis and the idea of design. The content mainly includes the following: setting up of the emotion dictionary, participle and labeling of lexical category, setting up of the Chinese dictionary, emotion modeling and recognition and the syntax analyzing of probabilistic context-free grammer. When participling and labeling of lexical category, the method of HMM model which uses the noting of part of speech is studied. The HMM is recognized as a very effective emotional modeling method when doing experiment for emotional modeling and recognizing. Especially, it improves the accuracy by using the emotional modeling method of embedded E-HMM. As using the sentence analysis of probabilistic context-free grammer, the algorithm of Inside and Outside which improves the accuracy of PCFG rules about setting probability parameter values is proved in this paper, thus realizing the system demand of real-time under the situation of assurance of recognition effect. At last, the validity of the Chinese character recognizing system is verified through the experiment.