Sentence sentiment analysis basedon syntactic construction of three-layers

The analysis on text emotional has attracted much attention as the explosive growth in data on the Internet. This paper proposed a method of sentence sentiment orientation based on syntactic construction which is quite different with the traditional methods of classification. It takes full account of logical relationship in sentences and establishes a sentence model of three layers according to the relationship between sentence elements. First, we use the sentiment word dictionary to score the words in sentences. Then we adopt different algorithms according to different layers and relationship. Finally, we obtain the sentiment score of the sentence by calculating the sores of the three layers. We apply this method to Weibo data and get a good result.