Sentiment analysis of Turkish political columns with transfer learning

Thesis Type: Postgraduate

Institution Of The Thesis: Orta Doğu Teknik Üniversitesi, Faculty of Engineering, Department of Computer Engineering, Turkey

Approval Date: 2013




Sentiment Analysis aims to determine the attitude (sense, emotion, opinion etc.) of a speaker or a writer with respect to a specified topic by automatically classifying a textual data. With the recent explosive growth of the social media content on theWeb, people post reviews of products on merchant sites and express their views about almost anything in their personal blogs, pages at social network sites like Facebook, Twitter, and Blogger. Therefore, sentiment analysis has become a major area of interest in the field of NLP. Up to date, most of the research carried on sentiment analysis was focused on highly subjective English short texts, such as product or movie reviews. In this thesis, sentiment classification is applied on Turkish political columns. Both sentiment analysis on news domain and Turkish language received less attention. Besides, in order to reduce the expense of collection and annotation of the data, Transfer Learning, which is recently adopted to sentiment classification tasks, mechanisms are applied to sentiment analysis of Turkish political columns.