Recognition Method for Emoticons Based on Network Text

摘要

Emoticons are widely used in the network text, which has enhanced effect for the author's emotional experience and semantic representation. Effective identification of the emoticons in the network text has great significance to sentiment analysis, public opinion survey and commercial investigation. Through analyzing the information about a number of emoticons on the Internet and based on the UTF-8 coding, new method of recognition emoticons is presented. First of all, the emoticons are differentiated and analyzed by using the statistical method, and then, the method describes the process by the steps of conditional information entropy are defined, computing mutual information and filter the candidates with using written rules. Finally, experimental results prove that the recognition algorithm has correctness, precise and practicality.