Exploiting Emoticons to Generate Emotional Dictionaries from Facebook Pages

During the first events of the Tunisian revolution, the social network, Facebook, played a key role in Tunisia and everywhere in the world. It became the first political tool that allows the Tunisian people to share trending news in actual time. Facebook provides the opportunity for users to comment on the news by expressing their sentiments. In this paper, we focus on emotion analysis of Tunisian Facebook pages. To do this, we first collect comments from the Facebook pages in order to analyze sentiments written in Tunisian dialect. Then, we propose a new method for emotional dictionaries construction. In fact, we distinguish nine emotional classes: surprised, satisfied, happy, gleeful, romantic, disappointed, sad, angry and disgusted. At this step, we focus on the use of emotion symbols as indicators of sentiment polarity. Finally, we present the experimental results of our method. Our system achieves effective and consistent results.