Sentiment value propagation for an integral sentiment dictionary based on commonsense knowledge

With the rise of social media, sentiment analysis has become a popular research field in recent years. A sentiment dictionary plays an important role in sentiment analysis. To remedy the lack of Chinese sentiment dictionary, researches translate English sentiment dictionaries to Chinese before using them in their applications. However, cultural difference often causes such translations inaccurate. Moreover, some of the dictionaries have small vocabulary size and some of the dictionaries have only polarity labels or even no label instead of value labels. In the first part of the paper, we integrated several common sentiment dictionaries to get the sentiment seeds. Then, we proposed a self-training sentiment spreading activation to expand the sentiment values on Chinese ConceptNet. Finally, we derived iSentiDictionary, a Chinese sentiment dictionary with 28,248 concepts and corresponding sentiment values.