About detecting the online style and messages - the point here is that we don't want DailyBot to be able to read all the messages that a person is sending through the chat, we believe it's important to respect the privacy and avoid processing all that content. Currently, DailyBot can only process direct messages between user and chatbot, or in channels when @DailyBot is mentioned.

A lot of it comes about because people are lazy: - unwilling to try something obvious/easily discovered - unwilling to think about who might actually be a sensible person to contact - unwilling to spend even a tiny effort to describe their problem clearly with basic useful details

These would be at the harder end of NLP but they are common and I suspect there are signs that could give away this sort of thing.

The good thing about flagging is that DailyBot won't need to process all the messages, only the ones that a person wants to flag.