Database exploration with dynamic abstractions

Abstract

For the efficient management of a large database we need a more friendly user-interface, more powerful abstraction capabilities, and more flexible data structures than is usually provided. A new database system called SCOPE was developed with the following three objectives: (1) unsophisticated end-users should be able to explore and navigate in a database system fairly intuitively, (2) the structures and the operations related to representing the data should be easily and efficiently modifiable, and (3) there should be abstraction techniques that allow the same data to be viewed in multiple ways. In the navigation for finding the desired data, views (or scopes) of the data are automatically generated with different levels of "coarseness" to facilitate the user's exploration of the database. Each scope is automatically and efficiently generated using certain dynamic abstractions of the data. Furthermore, an attribute-oriented tree structure is used in SCOPE to handle its data ( including token data and metadata) uniformly and flexibly. An operational prototype of SCOPE has also been implemented to test the design.