Social Media are at the heart of our communications and are among the most visited places on the web. Their user-generated content allows the gathering of immensely many information that cannot be processed entirely by the users. Thus it is of interest to understand how these users are deciding what pieces of information they are trusting and what are the reasons that influenced them. In this thesis proposal, we survey three papers presenting different methods to detect influence in social media. We show that the content produced is an important feature which hasn’t been extensively studied yet and deserves a closer attention. Toward this goal, we present our first results of influence detection via emotion recognition and propose several extensions.