Research Of Weibo Emotion Identification Based On OCC Model

Social networks now have become very popular and well-developed,and text is still one of the most commonly and most customary ways of expression.Weibo as a platform with active users contains a wealth of ideas and emotional information about things and events,this has important application value in the field of emotion detection.The traditional method of affective computation is relatively simple to use more emotion category,the most common method is dividing emotion appear in the text into three categories(positive,negative and neutral).However,this simple emotion classification system is not sufficient to summarize the text in rich emotion.Put this as a starting point,this paper research the multi-class emotion recognition of Chinese weibo.In this paper,we proposes a Flat-OCC model which summed 14 emotion categories combined with the characteristics of weibo corpus,compared to simple emotional categories,these 14 categories of emotion almost covering the emotional weibo appear,while ensuring each emotion has a certain number of experimental corpus.The main work of this paper include:1.Given the connotation and extension of the definition of each emotion,and further clarify the various elements of the rules of the determination of each emotion,and makes it easier to describe each emotion with formal language.2.Proposed multi-class emotion recognition method based on the Flat-OCC model,specifically,the model follows some module like these:preprocessing model,use the natural language processing-based tool for weibo corpus segmentation and content filtering;corpus tagging module,achieve a variety of emotional categories of corpus tagging,and detecting tagging effect through a visual method to ensure the strategy that we can always adjust the tags.With the more and more weibo get annotated,we get a certain amount of multi-class emotional corpus;emotion recognition module,this module uses two methods for emotion recognition,the first to use heuristic rules to recognize the emotional type corpus contained,meanwhile iterative update keywords tag phrase subsequent used,then use the keyword sequence tags to identify the emotional type again.Furthermore,keyword sequence tag as a fine emotion discrimination rules,can be used as a further supplement to the Flat-OCC model to improve the accuracy of determining the different categories of emotion,therefor improving the quality of multi-class emotional tagging corpus.3.Compemotion are the effect of heuristic rules and keyword sequence tags for multi-class recognition and their advantage/disadvantaged and for the environment by experimental verification analysis.