If you search for “world series,” chances are that you want to either find out something about buying tickets for the 2007 world series or learning something about the 2006 World Series. It’s less likely that you would be trying to find out about an older “world series.”
Some queries can be tied to specific events that way, and the people at search engines may try to understand which events people are more likely to be searching for when entering a phrase like “world series” into a search box.
A Microsoft patent application explores how selections in search results for certain queries may help a search engine cluster those results into different events that may be semantically similar, and close together in time.
Event detection based on evolution of click-through data
Invented by Tie-Yan Liu and Wei-Ying Ma
Assigned to Microsoft
US Patent Application 20070214115
Published September 13, 2007
Filed: March 13, 2006
Abstract
A method and system for detecting events based on query-page relationships is provided. The event detection system detects events by analyzing occurrences of query-page pairs generated from a user selecting the page of the pair from a search result for the query of the pair. The event detection system may identify semantic and temporal similarity between query-page pairs. The event detection system then identifies clusters of query-page pairs that are semantically and temporally similar.
It’s interesting – I came across this old update, and these days this sort of thing is fairly commonplace with respect to new developments like Google Instant and Google’s suggestion dropdowns.
Hi Tom,
It does seem like there’s something like this going on in Google these days as well, especially in places like query refinements and suggestions through Google’s dropdowns.