Competition To Identify Sexual Predators In Chat Logs 273
An anonymous reader writes "Researchers from the University of Lugano, Switzerland, and other universities from the U.S. and Europe organize a competition to automatically identify sexual predators in chat logs. The task is described as: 'The goal of this sub-task is to identify classes of authors, namely online predators. You will be given chat logs involving two (or more) people and have to determine who is the one trying to convince the other participants(s) to provide some sexual favor. You will also need to identify the particular conversation where the person exploits his bad behavior.' Their data set covers hundreds of chat logs with dozens of true positives (i.e., chats where one is trying to hit on another)."
Simple algorithm (Score:5, Interesting)
Re:False positives? (Score:5, Interesting)
Some friends of mine did that once, and put them into a teen-oriented IRC channel. "Lace" was about thirty lines long, and mostly just pretended to be teenage girl playing with anatomy and inviting others to do the same. The other one, whose name escapes me, was about a gig and a half from all of the conversation it had absorbed.
The big one was kickbanned quickly, within a few minutes. "Lace" remained in the channel for the better part of an hour until people stopped replying to it, and was finally booted when the moderator-bot kicked it when "Lace" posted, "you're boring! I shouldn't have left #hotstuff" or something like that, citing that "Lace" was advertising for other channels.
The guy who had written the serious, big one was devastated. The guy who wrote lace had the rage-comics troll face look for the next several days.