Today hundreds of researchers from Microsoft’s worldwide labs in China, England, India, and the United States will gather in Redmond to exchange ideas with colleagues, show off their latest projects, and provide a glimpse into the future of computing. There are some pretty cool search technologies being shown that we wanted to give some love to here on the Live Search team blog.
GeoLife 2.0: A Location-Based Social Network
GeoLife is a GPS-data-driven social network on Microsoft Virtual Earth. It is not only a Website where individuals can manage, visualize, and understand their life experience using their own GPS trajectories; it’s also a social networking service that enables people to build connections with each other based on location histories.
The Opinion Index in Live Search today is built off of research started in this project. The current project takes that work to the next level, collecting, storing, and organizing opinion data such as user reviews of computers, electronics, software, video games, restaurants, and hotels. That information will ultimately be used to help searchers more easily make informed purchase decisions.
Renlifang: Web-Scale Entity Summarization
Currently, information about a single entity (such as a person or a product) might appear on thousands of Web pages. Renlifang is a web-mining summarization system that extracts information about particular entities from billions of Web pages, reducing the number of pages a user has to comb through to find the information they are looking for.
Color-Structured Image Search
Color-Structure Image Search is a new image search interface that uses rough color layouts to indicate users’ intent (instead of using keywords only). This approach helps users find images that can be roughly described by color spatial distribution. For example, you may want images of a tiger on a sunny day, under a blue sky. If you search with the term “tiger,” it is difficult to find this type of image in the first few pages. If you expand your search term to “tiger sky,” it is even more difficult to find tiger-related images. With Color-Structured Image Search, we’re able to re-rank search results for the query “tiger” to include tiger-related images with “blue” color at the top of the search results.
Check out the Microsoft Research TechFest 2009 homepage for more information including video clips and demos of some of the most innovative work from our research labs around the world.
Harry Shum, Corporate Vice President of Search Development, Live Search