TL;DR: I'm running a project to detect low-quality posts on EL&U, but with a distinct lack of experience here I don't know what counts as a low-quality post. Help?
In the few days since Tim Post made some tweaks to the spam system, I've been running a fork of SmokeDetector in this chatroom to try to catch some of the abusive posting patterns EL&U has been seeing. It's caught a good number, and the number of incoming posts has slowly reduced.
However, SmokeDetector is a stop-gap. It was designed for spam, and the monkey-patches I made to apply it to this situation are... fragile, shall we say. It's nowhere near a permanent solution.
Which, together with an idea from tchrist, gave me the slightly questionable idea of starting up a machine-learning project to detect all low-quality questions on EL&U, not just this recent wave. Great, awesome, I've written half the code already.
Problem is, I don't have a whole lot of experience here on EL&U, so I don't actually know what constitutes a low-quality question here. This is where I need some help from more experienced people here - what's low-quality? For example, do questions that show no research effort count as low quality? Homework (do you even get any of that)? For the sake of this project, we'll define low-quality as things that should be closed and deleted, to make it easier to define.
What's low-quality here?