چكيده به لاتين
The rapid growth of social networks and media creates massive texts about people's opinions on various issues. For this reason, understanding people's opinions about different subjects are important in decision-making. Opinion mining by retrieving information and extracting knowledge from the text, using the techniques of data mining and natural language processing, analyzes and review the views of individuals. In other words, opinion mining, it means identifying the positive or negative opinions of people about a problem or product. In this thesis, Persian dialogues have been used for both Apple and Samsung phones. In the early stages of the research, we extracted the terms and words containing comments using the double propagation and improve the rules of this method. In the second phase, by using the aspects and words containing the opinion, the polarity of the aspects in the sentences is discussed. In the third step, using the aspects and words containing the comments, and the role that the words have in the sentence, we summarize the comments. In the final step, summarized comments are classified using support vector machines and naïve bayes.The results obtained from both methods are comparized between complete and summarized sentences.The results indicate that
the proposed methods have yielded good results.
Keywords: Opinion Mining, Feature Extraction, double propagation algorithm , Support Vector Machine algorithm, Naïve Bayes
