QueryCat is a new site that indexes FAQs and makes them searchable via ordinary language queries. It claims to have indexed over 4 million FAQ questions.

Cool idea. FAQs have enough predictable structure that the questions can probably be pretty cleanly separated from the answers. (For one thing, them question thingies tend to end with curly marks.) And, of course, the info in FAQs is, by definition (well, if FAQs actually compiled questions that were frequently asked) frequently requested and thus valuable.

I did a little poking around. “Where can I get a free blog?” got 664 results. ” “How do I tune up a bicycle?” got 11 hits. Where can I get vegetarian omega 3?” pulled up three answers, all pretty relevant. “Should I reply to spam?” got 372 hits, some generic and some specific to particular mail programs or sites. “Does putting in a new hard drive invalidate the warranty on my thinkpad” and “What’s a normal triglycerides result?” got zero hits. So did “What movies has Lily Tomlin starred in?” because that’s unlikely to be asked in a FAQ.

As is common for sites that let you type in questions, “How do I tune up a bicycle” gets exactly the same results as “tune up bicycle” (all without quotes). Natural language = no stop words. Who cares, if it works? It’s also not very forgiving of misspellings and variants; remove the space in “omega 3” and you get zero results. “Fix a dent in my car” gets zero hits, “Repair a dent in my car” gets six, and just “dent in my car” gets 12.

I wonder if QueryCat keeps track of the context of the questions it indexes. So, if 9 out of 10 of the Q’s at the Acme TNT FAQ use the phrase “Acme TNT” in their replies, but the tenth Q&A — “Q: Does this blow up coyotes real good? A: Yup. Real good.” — does not, will that tenth Q&A show up at the top of the list when some queries “Can I blow up coyotes with Acme TNT?”? Just curious.

I did run into one anomaly, or maybe I’m just confused by the site’s UI. The results are listed in the usual search engine format, with a link to the URL, a description, and then some more links. But in my limited poking, the links led to the home page of the sites, even when the anchor text said it linked to the FAQ.

Anyway, it’s a nice idea for a site and could quite possibly be helpful, especially when your question uses terms that would open up the floodgates of normal search engines .

[Tags: querycat faq metadata everything_is_miscellaneous ]

Categories: Uncategorized dw