Microsoft hopes to harness some of the time and energy that consumers devote to games to solve computational search problems that otherwise are difficult for researchers to tackle, according to the authors of a new paper presented at last week’s SIGIR 2009 conference in Boston.
The goal of a new online game called Page Hunt is to elicit data from online players that “can be used to provide metadata for pages, provide query alterations for use in query refinement, and identify ranking issues,” wrote a Microsoft Research team led by Raman Chandrasekar and Chris Quirk.
A Test Spin
Microsoft Research has already launched an online demo version of Page Hunt that gives gaming aficionados an opportunity to take it for a test drive. Here’s how it works:
Players are shown a random Web page without the URL and are asked to submit one or more words as the most likely query for the site. The game obtains the top search results for the query from Bing and displays them in order. Points are awarded to players landing in the top five query slots, with 300 points being the top score per search.
The primary goal of Microsoft’s Page Hunt effort is to generate Web user responses that search-engine designers can use to fine-tune their search algorithms in ways that best mirror the human search experience. At each step of the game, the researchers record the player’s screen name or an anonymous id; the Web page URL; the query; whether it was successful and, if so, at what rank position; the time; and the points the player got for this query.
“The player’s query behaves as a tag or label for the page,” the report’s authors observed. “When the player gets it right, this is valuable; but even when it is wrong,…