The Research On The Sentiment Orietation Of The Text

The Sentiment orientation analysis of the text is a branch of the studies on thefrontier computer field, the intelligent information field and natural languageprocessing field.The commented object can be judged with comments of subject’s thatthe sentiment orientation of the comments is commendatory or derogatory. Sentimentorientation of the text is widely used in information retrieval, information filtering,sentiment recognition and other fields.Particle size according to the size of the text can be divided into three levels: words,sentences, text. The methods used on the three levels sentiment orientation haveconnections and differences. The paper starts to study the most important and the mostbasic words, which also achieve the sentiment orientation analysis of the sentence andtext based on the words. The main research work and results are:First, the sentiment orientation analysis of words: This word-based semanticsimilarity algorithm which has been improved, through the test, the improved algorithmcan increase the accuracy of the semantic dictionary.Secondly, the sentiment orientation analysis of the sentence and phrase: This paperis based on natural language processing and the dependencies between the words, then Iextract the modifying adverbs and the sentiment words from a phrases or a sentence,and I make use of the SO-PMI algorithm to compute the polarity of the sentence and thesentiment orientation of the sentence. In the experiment, the Key semantic words areextracted using a new algorithm, the Key semantic words has more extensive coverage.The new Key semantic words consist of the general area and the specific areas.Finally, the sentiment orientation analysis of the text: The method of traditional textclassification is used for the sentiment orientation of the text in the paper. Theexperiment will extract the sentiment words from the text as the features of text, usingthe SVM classifier to achieve the classification of the text.