The Research On Analyzing Affective Tendency Of Text

Now it is more and more active in the area of natural language processing about analyzing the affective factors in the text, the purpose is to evaluate a judgment for objects held in a article is to support or oppose. It can be widely used in the text retrieval, text filtering, and other fields, and this also gets more and more attention from various aspects. To quantify the affective of the text, a lot of emotional data can be provided, which led to more on the emotional aspects of the application. The main research work and achievements are:1. This paper focuses on the algorithm of the text tendentious based on semantics, which assess the emotional tendencies of the Chinese text and based on natural language understanding. The algorithm through partial sentences tends to express emotions and evaluation of the word-object combination of the phrase, in order to replace the grammar's structure of the sentence and analysis its semantics. By testing, in the text emotional semantic feature extraction, using these phrases in these sentences, it can basically express the sentence emotional tendencies, and meet the classification requirements. And using this algorithm, the analysis of the temporal and spatial are more efficient, we can achieve practical application needs. Through experiments, the algorithm of articles on specific areas of investigation-classification rate can reach 89.8%, the recall rate can reach 93.5%.2. This paper proposed the text classification algorithm based on the analyzing affective tendencies, which focus on the words and phrases in the text. The algorithm is to analyze texts' emotional words and the phrases as the text's emotional characteristics, and use these characteristics to achieve the emotional text's discrimination. The algorithm improves the inadequate of emotional tendencies in the discriminate because of a lack of semantic constraints based on past statistics emotion of the algorithm.3. It proposes the algorithm of emotional tendency, and mainly includes the emotional vocabulary selection, emotional dictionary logo and the sentence structure of the text and emotional tendencies value judgments. Through segmentation,tagging,effective phrases extraction, computing the value of the emotional tendencies of the text and so on to achieve.