On NPR this morning John Gordon had a piece on how Microsoft and MIT are working to develop a more personalized search method.
The method considers the content of your e-mail, documents, blog content, and Internet history to allow personal interests to influence the final search results presented to the user.
The interviewee, M.I.T. doctoral candidate Jaime Teevan, claims that the Google and Yahoo personalized search tools are based on a lightweight user model because they use data that the user provides and the pages they’ve already visited. The Microsoft/MIT personalized search would take into account more than just the web pages recently visited to include other documents and information on a user’s computer and re-rank search results accordingly.
I think this work goes along with the notion that in order for search engines to evolve and improve search quality, they will need to do a better job at understanding the user’s intent. Including information from the computer used to conduct the search does make sense but only if that computer is not shared by others. Without considering different users, I can’t see how a system like this would work as well as it’s being positioned.
Listen to the Podcast here