Indexing Real-World Data Using Semi-Structured Documents. Information Systems

摘要

We address the problem of deriving meaningful semantic index information for a211u001emulti-media database using a semi-structured document model. We show how our 211u001eframework , called feature grammars, can be used to (1) exploit third-party 211u001einterpretation modules for real-world unstructured components, and (2) use 211u001econtext-free grammars to convert such poorly or unstructured input to semi-211u001estructured output. The basic idea is to enrich context-free grammars with special 211u001esymbols called detectors, which provide for the necessary structure just-in-time 211u001eto satisfy a parser look-ahead. A prototype implementation has been constructed 211u001ein the Acoi project to demonstrate the feasibility of this approach for indexing 211u001eboth images and audio documents.