From the Microsoft Research Labs in Redmond, Washington, there comes a new online game whose goal is to improve the algorithms powering Microsoft’s search engine Bing. The game, called Page Hunt, shows you various pages from around the web and then asks you to guess what queries would make Bing display that page within its first five search results. If the query you enter returns the page as the number one result, you’ll get 100 points, if it’s the number two result, you get 90 points, and so on. But basically, if the page is in the top five results, you win.
To make the game fun, Page Hunt includes timed responses, score keeping, a top-scorers list, taboo queries, and bonus points…like those received for avoiding frequently-used categories.
Page Hunt was developed by Chris Quirk and Raman Chandrasekar at Microsoft along with colleagues from Georgia Tech and was debuted at the recent SIGIR09 conference in Boston. Their research paper describing the system can be found here.
Already, the researchers have discovered some interesting information from the data collected: the longer a web page’s URL, the harder it is for users to match the page to query words.
They don’t speculate as to why that happens, but it demonstrates how the game can provide data that will uncover interesting insights into how search works.