Interoperability and Multimedia Archives

In distributed computing systems, it is unwise to move data to the point of program code, but instead process data at the point of storage. This concept is even more appropriate to multimedia repositories where large files must be processed as a result of each retrieval operation. Two problems with multimedia databases are that they require a large volume of disk storage and processing power and that it is generally difficult to query video content or optimise the retrieval process. In the EGTV project, both of these issues are addressed. In the first case, the data server is distributed across multiple autonomous sites where users are permitted to modify schemas and in some cases the data model. In the second case, a mechanism for storing behaviour has been devised which allows functions such as retrieve_frame, retrieve_segment and retrieve_context to be implemented and stored at individual servers. This provides selective retrieval and thus, large data items can be processed locally to filter unwanted data before the transfer process must begin. In this way, storage of operations forms the basis for optimisation of retrieval content and volumes.