Activities

Many text databases on the web are ``hidden'' behind search interfaces, and their documents are only accessible through querying. Search engines typically ignore the contents of such search-only databases. Gravano and Ipeirotis , in collaboration with Mehran Sahami, from E.piphany Inc., have developed a novel strategy to automate the classification of search-only text databases. Their technique starts by training a rule-based document classifier, and then uses the classifier's rules to generate probing queries. The queries are sent to the text databases, which are then classified based on the number of matches they produce for each query. A paper describing initial results for this problem was presented in May at the WebDB'00 workshop. (See reference below.) Also, a demo system showcasing the main aspects of this research work was developed and presented at the June all-PI DLI2 meeting in the United Kingdom.

Gravano, Ipeirotis, and Sahami are currently further developing the database categorization work, including a large-scale evaluation of the new techniques over web-accessible databases. The goal is to complete the development and evaluation of the categorization techniques by mid fall. Additional plans for the summer and fall include the exploration of distributed search protocols to facilitate metasearching, and the initial deployment of a small-scale metasearcher that will provide access to a number of web-accessible text databases.

In addition to searching text, we are also exploring techniques for storing, indexing and accessing echocardiogram video. Echocardiogram video is a popular diagnosis technique used in cardiology. The intended users include physicians, medical students, as well as patients. We have designed the components for storing, indexing and random access of echocardiogram videos.