Research On Personal Emotion State Based On The Text

In the fast pace era, people have to face more and more pressure. In order to relieve the stress, some people find relatives or friends while others write diary. With the development of network and communication tools, more people would like to choose communicate on network with their opinions or emotions. The personal preferences are obtained by analysis these information so that network marking, public opinion monitoring, psychological guidance and so on can be implemented. The current research of emotion focused on images, sounds, behavioral and physiological signals. However, text sentiment analysis research is aimed at product reviews, public opinion supervision and so on, little research on individual’s emotional state, has not yet appeared more effective results.Text sentiment analysis is mainly used to distinguish text emotional tendencies, i.e. positive, negative or neutral. In this paper, the personal emotion state is studied based on the traditional text sentiment analysis. Firstly, the existing resources of emotional words are summarized and systemized. The emotional words are extended considering the characteristics of army and network. At the same time, the dictionary is constructed including the commendatory terms and derogatory terms. The transitional words, degree adverb words, negative words and so on which affects the tendency and strength of text emotions are summarized, especially, the extreme words(phrase)dictionary are constructed through communicating with the relevant departments. Then, texts are preprocessed, and emotional words in text are used as feature selection approach. The transitional words, degree adverb words, negative words, rhetorical question and so on are analyzed and processed accordingly. The emotional tendencies of sentences are calculated and the whole text emotional tendencies are obtained by comparing the number of mixed sentences. Finally, whether the text authors are needed to pay attention is obtained by weighting and threshold value. The experiment results show that the text emotional analysis method is improved by this paper.