Interactive Keyword Search for Relational Databases

The success of web search engines has made keyword search the most popular search paradigm. With the rapid growth of structured data repositories, supporting keyword search over such repositories enables users to pose queries easily without the need to have full knowledge of the database schemas or structured query languages. However, keyword queries are ambiguous and the challenge to retrieve answers that satisfy users' search intention remains.

In this project, we adopt a semantic approach to relational keyword search. We find that the interpretation of user's search intention is important and requires knowledge about the semantics of objects and relationships in the database. We introduce the notion of query patterns to depict possible search intentions and annotate answers to facilitate user understanding. We have developed a prototype system called PowerQ that interactively guides users to obtain relevant answers to their queries.

Selected Publications