Opinions Mining in Facebook

Abstract

Because of the tremendous development in all areas of scientific, economic, social and others, the need emerged to find unconventional ways in which we deal with text data that have become very large sizes these days.Therefore, we must find new ways to derive knowledge and hidden information within these huge amounts of data, and these techniques in exploration data that we used in this study are classification techniques.We fetch the data from Facebook social networking site, and then we have worked on cleaning and preparing the text to texts classification process. These texts are contained a lot of noise and information is useful for the opinions  analyzing process, such as advertisements, links, e-mail addresses, and the presence of many words that do not effect on the general orientation of the text.After we get all the posts from the Facebook page and its related comments, our goal is to know the Percentage of positives and negatives for this post.We applied Naïve Bayes algorithm in classification, we had the appropriate training, and after passing Posts and comments data (opinions), we got good results on the ratio of positives and negatives of the.

Keywords: Opinions mining, Sentiment classification, Challenge, Access token