Since I'm basically done with my natural language processing project this quarter, I thought I would write down an idea I have for my project next quarter for practicum.
Either I would do the ToA Constructor, or I would write something which can extract the structure of a document.
As it appears it me is should be simple. The first thing to try is to grab the HTML header tags, which would of course indicate a title. I don't think that will work on too many pages though, since not a lot of people like them. The next thing would be grabbing everything under a certain word length that's on a line by its own. I can think of them being either headings, or one liners made for effect. That's basically it. If these two don't work, I suppose we could also try scanning for a larger font size, but that takes in account CSS and all the other SGML junk, and is a lot harder to do.
Although, somehow I don't think I'll be working on this next quarter. But it's an idea.