System for collecting, transforming and managing media metadata

Abstract

An extensible framework for the automatic extraction and transformation of metadata into logical annotations. Metadata imbedded within a media file is extracted by a type-specific parsing module which is loaded and executed based on the mimetype of the media file being described. A content processor extracts information, typically in the form of time-based samples, from the media content. An auxiliary processing step is performed to collect additional metadata describing the media file from sources external to the file. All of the metadata thus collected is combined into a set of logical annotations, which may be supplemented by summary data generated from the metadata already collected. The annotations are then formatted into a standardized form, preferably XML, which is then mapped into a database schema. The database object also stores the source XML data as well as the original media file in addition to the annotation metadata. The system provides unified metadata repositories, which can then be used for indexing and searching.

Dependencies are inferred from the claim text.