Systems designed to help users answer questions have traditionally focused on analyzing formal content (e.g., full web pages and news articles) to find answers (or nuggets) to asked free-text questions. However, little attention has been given in building those systems to analyzing online informal content (such as millions of posts and tweets that are created daily on Facebook and Twitter). The rapid increase of popularity and interest in that type of media, especially in the Arab world, as both conversational and information dissemination channels, makes it a potential rich source of answers to real-time questions. In this proposal, we plan to address the problem of answering users’ questions from Arabic content in social media. While the type of data allows new user-centric questions to be asked (e.g., what are the different opinions on a decision made by a national figure right after it was made), it also opens up new challenges, such as dealing with different dialects, mixed languages, and conversational content, in addition to the unique characteristics of the Arabic language. We propose to explore the solution space from several different perspectives, e.g., ranked retrieval, topic modeling, and information visualization. In solving the problem, we plan to build a scalable open-source real-time system that answers given questions while providing plausible explanations of selecting the presented answers.