Automatic synthesis of semantic information from multimedia documents

DETAILED DESCRIPTION The present invention is a system for automatically synthesizing semantic information units from raw material extracted from non-textual documents.
It supports error detection and correction, allows mapping from pieces of raw information extracted from the media into semantic information entities in the application domain and provides a complete procedure for building up hierarchical information structures for non-textual media.
It adopts a formal specification to describe the synthesis process of constructing semantic media structures for all different media in multimedia documentation.
With this system, non-textual documents can be further processed and their content can be accessible and represented in a similar format to that of SGML structures from textual documents.
This provides single unique semantically-rich information structures to the hyperlinking process in multimedia applications.
The present invention includes an error corrector for receiving raw AIU's and for providing corrected raw AIU's.
A primitive identifier receives the corrected raw AIU's and provides primitives.
A basic synthesizer receives primitives and provides basic objects.
An abstract synthesizer receives basic objects and provides abstract objects which are in condition for the hyperlinking process.