Sentiment Computing of Web Financial Information Based on the Part-of-Speech Tagging and Dependency Parsing

Sentiment orientation analysis has attracted a great deal of attention recently due to many practical applications and challenging research problems. Traditional text sentiment analysis method based on bag of words model does not take into account the syntactic structure of the sentence, while the method of text sentiment analysis based on dependency parsing model tries to solve this problem. At present, the existed methods based on dependency parsing mainly focus on the author’s observation and lead to be subjectivity and arbitrariness when selecting dependency pair. Therefore, this paper firstly finds out 4 kinds of parts of speech which could affect sentence sentiment, such as adjectives, verbs, adverbs and nouns, from the original polarity, modified polarity and dynamic polarity of emotion words. Secondly, we analyze the effects of 24 kinds of dependency pair on the sentence sentiment computation and select 8 kinds of dependency pair according to the part of speech and Chinese modification understanding. Thirdly, 6 kinds of sentiment computing rules are designed from the combination of the part of speech of the above 8 kinds of dependency pair and then the sentiment computation method based on two binary tree is proposed, which includs the construction method and the sentiment computation method. Finally, the experiments are carried out on the Web financial information and the results prove the effectiveness of this method.