The mission of our group is to develop new technologies that enhance user expereince in Internet search and services. The members of our group come from diverse background, and combine their skills to evaluate ideas and rapidly prototype concepts towards realizing that goal.
Improving Search Quality
Search engines collect vast quantities of data in the form of queries, clicks, web pages, feeds, etc. This content is rich in terms of structure, user behavior and trends. By leveraging a massively distributed environment, we develop techniques to mine this data to improve search quality.
The aim is not just to build stand-alone applications, but rather to create an infrastructure upon which these applications can be built and deployed at Internet scale. Several of our projects involve cross-group collaborations, spanning both Research and Product groups.
The Web is abundant with ever-growing information which is inherently partially structured and heterogeneous in nature. Being a central part of cultural, educational, commercial and social life, the Web serves on a daily basis as an invaluable source of knowledge for 100's of millions of users looking for information on a multitude of topics; reading the news, writing about their experiences in blogs, researching for products and services, etc.
Our goal in this project is to provide Live Search users with efficient, transparent and effective Web search access to a variety of semi-structured and unstructured data sources on the Web. Our aim is to complement and enhance the users’ search tasks be it researching for a digital camera, identifying great educational sources for a given topic or determining a good practitioner to visit.
Our vision is to provide users with transparent, real-time and contextual access throughout their search experience on commerce, health and education.
- Generating Related Keywords: keyword generation is an important problem in web search that has many applications including query suggestion and topic classification. We have developed novel techniques for this problem and some of our research has been included in Live Search.
- Query Pattern Identification: user queries to a search engine often falls into certain patterns. We have developed automated techniques for identifying such patterns and associating these patterns with certain classes of search results. This has since been adopted in Live Search.
- Rakesh Agrawal will be serving as the Awards Chair for SIGMOD 2012.
- Ariel Fuxman served as the Publicity co-chair for SIGMOD 2011.
- Rakesh Agrawal received 2010 IIT-Roorkee Distinguished Alumni Award.
- Krishnaram Kenthapadi, Nina Mishra, and Alex Ntoulas's paper titled Releasing Search Queries and Clicks Privately (co-authored with Aleksandra Korolova) is one of six papers nominated for best paper in WWW 2009. Here are some links to press articles that discuss this paper:
- Making the Web More User-Friendly (Microsoft Research).
- Microsoft: How to study search data without risking privacy (TechFlash).
- Anitha Kannan has received the CVPR 2008 best student paper runner-up for her paper titled Epitomic location recognition (co-authored with K. Ni, A. Criminisi and J. Winn).
- Sreenivas Gollapudi has received the PODS 2008 best paper award for his paper titled Estimating PageRank on Graph Streams (coauthored with Atish Das Sarma and Rina Panigrahy).
- Ariel Fuxman has received the SIGMOD 2008 best dissertation award for his thesis titled Efficient Query Processing over Inconsistent Databases.
- Rakesh Agrawal has received the ICDE 2008 influential paper award for his 1995 paper titled Mining Sequential Patterns (coauthored with R. Srikant).