Retrieval activities in a database consisting of heterogeneous collections of structured text

摘要

The first part of this paper briefly describes a mathematical framework (called the containment model) that provides the operations and data structures for a text dominated database with a hierarchical structure. The database is considered to be a hierarchical collection of continuous extents each extent being a word, word phrase, text element or non-text element. The filter operations making up a search command are expressed in terms of containment criteria that specify whether a contiguous extent will be selected or rejected during a search. This formalism, comprised of the mathematical framework and its associated language, defines a conceptual layer upon which we can construct a well-defined higher level layer, specifically the user interface that serves to provide a level of functionality that is closer to the needs of the user and the application domain.

With the conceptual layer established, we go on to describe the design and implementation of a versatile interface which handles queries that search and navigate a heterogeneous collection of structured documents. Interface functionality is provided by a set of "worker" modules supported by an "environment" that is the same for all interfaces. The interface environment allows a worker to communicate with the underlying text retrieval engine using a well-defined command protocol that is based on a small set of filter operators. The overall design emphasizes: a) interface flexibility for a variety of search and browsing capabilities, b) the modular independence of the interface with respect to its underlying retrieval engine, and c) the advantages to be accrued by defining retrieval commands using operators that are part of a text algebra that provides a sound theoretical foundation for the database.