Detect online conversations which turns ugly

To avoid the main challenges of online social systems such as harassment and personal attacks researchers have found an algorithm of predicting from the initial of a conversation whether it will get out of hand. when an undesirable way of conversation is identified, the task aims to enable early, actionable prediction at a time will be made when the conversation might still be rescued. To this they develop a framework for capturing pragmatic devices such as politeness strategies and rhetorical prompts used to start a conversation, and analyze their relation to its future approach. This is really advantageous of detecting early warning signs of antisocial behavior in online discussions.

Share you ideas regarding this undesirable conversation rescuer algorithm.