The authors investigate possible improvements in online fraud detection based on information about users and their interactions. They develop, apply, and evaluate their methods in the context of Skype. Specifically, in Skype, they aim to provide tools that identify fraudsters that have eluded the first line of detection systems and have been active for months. Their approach to automation is based on machine learning methods. They rely on a variety of features present in the data, including static user profiles, dynamic product usage, local social behavior, and global social features.