Compression has been the sole objective of traditional audio and visual data coding schemes. More recently the requirements for effective information management have focused attention on the question of what mechanisms are required to support content based retrieval. This has been largely approached as a separate problem by either using index files containing meta-data or attempting to exploit whatever information is readily available in existing compressed data streams. This paper will review existing support for information management in current coding models, and contrast this with the requirements for effective retrieval. Additionally, a class of semiotically based coding models will be presented which are equally suited for achieving both high compression and effective information management.