baning is not the solution. we can devise a mechanism to connect social networking sites to cyberbullying by:
1) finding out a way (an algorithm) how different users are connected (related) to a particular user. this can be done after masking the real names. we can arrive at some figures, such as user1 is connected to 10 people as classmate (let's call it group1), 15 as schoolmate (group2) and 2 as sibling (group3).
2) now on a given day, if we see a spurt of messages (again, masked) from group1 to user1, with some expletives then these messages need to be looked into deeply and maybe, authorities need to be reported.
3) large number of messages can be ignored if sent around user1's birthday/anniversary etc and other important days, such as independence day/new year's day.
4) similarly, how many responses a video message gets or from which part of the world can also give some heads up for a potential case of cyberbullying.
5) messages posted at odd times from a cellphone can be treated as troublesome and may possibly help in avoiding a suicide bid.
all of this stuff can be done without knowing a person's name or location. only strong cases need to be explored and the real identities can be unmasked keeping the authorities (police, parents etc) in loop.
there are several ideas or techniques... don't know in which direction you need to go!
