Abstract

Study on network language viewpoint extraction based on affective computing

Author(s): Haiyan Gao

In recent years, the frequent occurrence of public sentiment has caught attention. In order to effectively control the incident, the study accurately extracted public views. The diverse content mainly includes the four main elements of public opinion (the theme, the holder, statement of holder and emotional tendencies of holder). By analyzing posts on internet forum, this study found that emotion tendencies are the key points of analysis. It is mainly about the qualitative analysis of tendency on post contents in forums. The content on the web forum discussion is varied with a variety of topics and witty cyber words. For the characteristics of this network language, the study manages to improve polarity dictionary by the extension of cyber word dictionaries, mainly to add some words used in network language experiments. The study attached great importance on the special words and special symbolsÂÂ’ influences in sentences, increasing the accuracy of the statement propensity analysis. An example is set to illustrate the problem of network language propensity: Content affective computing of theme A and theme B, if the article support A and oppose B, it means positive; otherwise it means negative. The study apples the above rules for affective computing, meanwhile statistical analyzes peoplesÂÂ’ emotional tendencies. This research achievement has great practical values.