I have a new paper up on First Monday. Maybe you should read it, if you’re interested in Q&A sites and the types of things people ask about.
Question types in social Q&A sites
by F. Maxwell Harper, Joseph Weinberg, John Logie, and Joseph A. Konstan
This paper is a collaboration between me and my advisor (HCI researchers) and two cool dudes from the humanities, in the field of writing studies (formerly: rhetoric). As it turns out, computer scientists and rhetoricians tend to think very differently about ways of contributing research to the world! This paper represents an early attempt to fuse some old skool Aristotelian rhetorical theory with some new school data mining.
Really, this paper reflects my personal frustration with much of the literature investigating user behavior on Q&A sites (like Yahoo Answers and Ask Metafilter). Most of the lit conveniently ignores the fact that there are a bunch of different types of questions that people are asking in online forums, and this fundamentally changes how we should view users’ intentions. It is indeed easier to assume that all questions are factual, and that “best answers” go to the best-written and most correct response. But these assumptions will lead to system designs that ignore lots of users.
This paper takes a shot at a more formal taxonomy of question types. I hope you find it useful or interesting.
Photo by matthileo
Ah, another inspiring example of irritation-based research!
I, too, have experienced frustrations with Q&A sites, but have found so little value at such sites, I generally tend to avoid them. Lately, I’ve been getting a stream of email notifications that various people are “following” me on Quora, which appears to be the latest Q&A hotspot. While a brief perusal suggests the signal-to-noise ratio appears to be better there, I appear to find it easier than many of my Twitter and/or Facebook friends – from whence I imagine these “following” messages are appearing – to resist the temptation to indulge myself in questioning or answering in that forum.
FWIW, my PhD thesis advisor, Wendy Lehnert, did her PhD thesis on question answering. I don’t know how easy it would be to find a copy (it was published as a book), but I found an online reprint of a couple of reviews of The Process of Question Answering. She employed a symbolic AI vs. statistical approach to the problem(s), but it may be of some interest / use.
Re: frustration. I understand. I think most of us in the field of computing tend to view things in terms of signal-to-noise ratios. And, I /really/ like Stack Overflow, accordingly. It turns out that others are less interested in such ratios. Many people use social software for deeply personal advice, opinion polling, and just reaching out to say “hey, is there anyone else out there like me?”. Crazy, but really interesting.
Re: Automated Question Answering. Thanks for the pointer. Turns out that there’s a lot of research still coming out in that area. Disambiguating the acronyms “Q&A”, “CQA”, and “QA” for the various, somewhat related research fields is tedious