X
Business

Microsoft Research shows off what's next for search

New Microsoft Research (MSR) search techniques and technologies under development were a hot topic at the March 6 kick-off of Microsoft's TechFest 2007 research showcase. On the short list: "Mix", a new search-based authoring tool, machine-learning technologies and new analysis tools desiged to improve search relevance.
Written by Mary Jo Foley, Senior Contributing Editor

New Microsoft Research (MSR) search techniques and technologies under development were a hot topic at the March 6 kick-off of Microsoft's TechFest 2007 research showcase.

(TechFest is open to press and analysts on Tuesday and to Microsoft employees on Wednesday and Thursday. Microsoft broadcast the opening TechFest address via a Webcast on Tuesday.)

Lili Cheng, a Microsoft researcher who helped Microsoft develop the desktop search feature that is built into Windows Vista, demonstrated a new "search-based authoring" tool known as Mix. Mix is a new user interface designed to help users aggregate, publish and share search results. With Mix, users can assemble "dynamic documents" that include search information garnered from desktops, intranets and the Web -- all the while maintaining appropriate privacy controls.

Cheng, one of the leads on Microsoft's "Wallop" research project, has been pioneering new ways to combine seemingly disparate social-networking software, such as wikis, RSS feeds and blogs, into a more unified and useful user experience.

Rico Malvar, the newly appointed head of Microsoft's Redmond, Wash.-based MSR lab, focused on how machine-learning principles can help customize contextually search results. Malvar described Microsoft's view of "evolving, personalized search," where search engines will be able to better "disambiguate things based on what you are working on."

Malvar showed off "Pictures of Search Relevance" technologies under development by MSR. MSR is investigating "key subsets of the Web graph" -- elements such as number of nodes and edges, graph diameter, connected components -- to help predict user behavior on search-related tasks. Over time, Malvar said, this data can be used to teach search engines to change result orders and better position ads.

Rick Rashid, the senior vice president in charge of MSR worldwide, emphasized in his opening remarks that Microsoft looks at TechFest as one of the key ways that MSR-developed technologies get transferred into Microsoft's product groups. Microsoft employees touring the TechFest booths this week may never have seen or heard of some of the projects under development by Microsoft's 750 researchers located throughout the world.

Rashid also emphasized that MSR can help the Microsoft product teams turn on a dime if and when new competitive forces come to the fore.

"If the world changes, we'll be able to change rapidly" thanks to MSR's work, Rashid told TechFest keynote attendees. "Chances are good" that Microsoft will have the IP portfolio, people and physical capital required to move nto new and emerging markets more quickly than might be possible otherwise.

Other new MSR projects that Rashid and his team showed off during the opening TechFest keynote:

* Epigraph: A kitchen display to help families monitor family members' whereabouts and activities using "remote presence" techniques;

* BubbleBoard: A visual answering machine

* Asirra: A system that initially classifies images of dogs and cats, but ultimately could help discern real human input from automated spam scripts;

* Boku: a lightweight programming model designed specifically for kids using Xbox 360 game controllers; and

* Community Buzz, a tool developed jointly by MSR and Live Labs, that combines text mining, visualization techniques and thread analysis aimed at online community participants, advertisers and product managers. (Community Buzz seems to be an outgrowth of MSR's Netscan project.)

Editorial standards