Social networks have become by far the most important place online to discuss and exchange information, but they also tend to polarize discussions and isolate users in their own cultural or ideological bubbles. This project aims to develop new methods to analyze the structure and mechanisms of influence within these communities, and to estimate the overall activity of social networks from only a partial view of them. Based on these methods, a tool will be developed to enable journalists to monitor the propagation of controversies and misinformation on platforms such as Twitter, YouTube and Reddit.