'The Guardian' published two extremely interesting articles about an analysis it made on 70m comments to its articles since 1999.
In the first article, they talk about the main results of their analysis. It is no surprise that they found out that the majority of abusive comments are targeted at article authors who are women and minorities. They also discuss which sections and topics draw the majority of the nastyness from the online crowds.
In the second article talks about the technologies and methods behind the analysis. It is impressive (and a bit inspired) the vast number of techniques and systems that must be employed to execute even the simplest descriptive analysis of such a large corpus of data.
I think that the analysis of online abuse deserves much higher attention from the scientific community. An extremely large number of people are connected, and for many of those, the internet is the platform that gives them their voice. The Guardian article makes it clear that unfiltered abuse creates a chilling effect that prevent many from being able to effectively participate in cyberspace.