What Time is it at Google Headquarters
November 9th, 2007 · by David Bradley
It is 10h45 GMT on November 8, 2007 as I type the first draft of this post, but you probably aren’t reading it till much later. It is not that it is going to take me that long to write a publishable version, it is just that the subject matter Cambridge company True Knowledge may be more relevant to a wider audience further down the line.
Anyway at the time of writing the company had just released a limited private beta test of its natural language search engine. What’s natural language? I hear some of you ask, others are way ahead and thinking “Google Buster!!!”. Well, a natural language query is more akin to what real people ask each other when they want to find information rather than the stilted, often Boolean based, questions we usually ask of search engines.
So, for instance, “What time is it at Google headquarters?” whereas “time AND google AND headquarters” would be something like a typical Boolean query you might use in advanced mode in a search engine. The latter takes time to construct, you might need to do something like this “time AND (google OR yahoo) AND headquarters NOT uk” for example, whereas a natural language query would simple be “What time is it at google or yahoo headquarters outside the UK?”
You get the idea. Hopefully, True Language will enter public beta fairly soon so you can test it yourself, but at the time of writing, most users are stuck with a few nifty screenshots and some explanation of how natural language queries work and how the system comes up with the most likely answer to that pressing question and a form to fill if they want to get recruited for the private beta.
The video below reveals some of the possibilities of True Language’s technology.
Oh, by the way, at the time of writing, it’s 01h45 PDT on November 8 at Googleplex, Silicon Valley, but not for long.


















1 response so far ↓
At the search engine level, this kind of natural language parsing is a formidable challenge. Not something for the faint of heart .. or a PHP developer, like myself!
If I have to write a search tool for the web applications I develop, I often opt for a cheap & cheerful natural language search engine.
So for example, the client can search for: “all Bookings by WayneSmallman from 11/2/2005 to 11/12/2007″, which is grabbed and then filtered by a battery of regular expressions, which then pull up the appropriate class methods to generate the search results.
It’s worth the effort because the power you’re handing to your clients is massive.
However, even relatively simple queries like “when is the next high tide in Maui” are probably out of my league…
Leave a Comment