Sentiment detection for predicting altruistic behaviors in Social Web: a case study

With the advent of Social Web, the user has become an active consumer which shares information and participates in social networks, online communities, blogs, wikis, feeds and chats. The volunteer person-power is a valuable resource which creates innovative content and helps other users to make right decisions with his own opinions, suggestions, advice. Opinions and suggestions have an amazing impact on the online user community: they may unexpectedly influence decision-making activities starting from simply buying or not a smartphone until to social events, political actions, and even marketing strategies. This paper aims at studying the role played by the sentiments in influencing the user actions. Particularly, it analyzes how sentiments expressed in the text can move the reader to do altruistic actions. The idea is from RAOP community where users write posts asking or offering a free pizza. Our work achieves a comparative analysis of machine learning methods on a ROAP dataset, that collects original posts where users asked for a free pizza. The goal is to extract the sentiments expressed in natural language, in the textual requests, in order to predict which user request will be satisfied (getting a free pizza). Finally, a posteriori "affective" analysis shows the predominant emotions expressed in the satisfied requests, that move the readers to have an altruistic behavior.