TITLE: ``Intelligent Query Processing over Complex Databases''

ABSTRACT:

Formulating queries over complex databases can be a difficult task requiring extensive syntactic and semantic knowledge. Syntactic knowledge is needed to ensure that a query is well-formed and references existing relations and attributes. Semantic knowledge is needed to ensure that a query satisfies user intent. While a user often has a general understanding of the contents of the database, design decisions to structure and store data in optimal ways may conflict with the user's intuition or expectations. Consequently, database specialists are needed to formulate queries, and ad hoc access to the database is impeded.

This presentation describes a research project in automated query formulation that significantly reduces the amount of knowledge required by database users. In particular, an extended Entity-Relationship model is employed to formulate queries automatically in response to high-level requests that specify only attributes of interest. By using knowledge of the target query language, syntactically correct queries are assured. Similarly, by using a knowledge-rich conceptual model of the database, the most likely intent of a request can be inferred and generated queries can be semantically optimized.

If time permits, a demonstration will be given of a prototype system known as QUICK. In addition to facilitating research, QUICK has been used (1) as a design tool and query formulation module of an intelligent interface developed by the Space Telescope Science Institute for managing complex databases, and (2) as a platform to explore data fusion among logically disjoint databases by the U.S. Army Research Laboratory.

Everyone is welcome to attend.

Web Accessibility