Interoperability between video libraries is valuable because it helps to build collections through combinations of related archives. In addition, it helps to distribute the costs and time required to build large bodies of information.

Accomplishments

CMU/Informedia and MIT/INTERNET CNN Newsroom (managed by CAES/MIT) cooperated to examine issues pertaining to interoperability by performing an experiment to make the Informedia Library data available to various clients as well as making the Informedia client able to access libraries other than their own. The activity was designed to establish how two different video libraries would interoperate with the constraint that the data organization and client applications of each would remain unchanged. Each client was to discover and navigate each other's content, and then retrieve each others' Objects (metadata and media playback). In addition, the project was to demonstrate interoperability among distributed remote repositories (WAN) as well as a local repositories (LAN). The specific criteria for the Interoperable prototype system included:

1. Search one another's sites.
2. Play one another's content.
3. Pay one another's charges.
4. Respect one another's special datatypes and reserved functions.
5. Negotiate for available services.

Figure 4: Illustration of the Demonstration Interoperability Architecture

Interoperability was successfully achieved between the Informedia Digital Video Library and NMIS Library and client which implements streamed video via a Web Interface. The NMIS library content consisted of INTERNET CNN Newsroom clips, a daily summary of important news for educational use in grades 7-12. Informedia library content consisted of documentaries about the war in Bosnia and various science archival materials. The study established that a good approach to solving many interoperability problems (with the constraint of not significantly changing libraries or clients to adhere to the same standards) would be to build intermediate filters or agents which translate from each client's known datatypes and requests to/from each library server's known set of datatypes and services. The final deliverable implemented this type of "mediation" level in the areas of metadata translations, cache management, and object name-space resolution [13].

Future Directions

Possible future directions for further interoperability work were identified. These included developing a basic Object Model to share at a high level between Digital Libraries. For instance, there could be a high level set of metadata for each item in the catalog, i.e. "Title", "Author", "Abstract", etc. A high level protocol was also proposed whereby a client could query and find out what type of data and what type of services a library exports so that library services could be negotiated much like standard network services like telnet. Finally, further work was suggested for creating global naming and (object) location services.

Due to this work, the Informedia project began a joint effort with ISI and ARPA to apply Informedia technology to content provided by ARPA. The goal of this work was to seed their own video library as well as to showcase the Informedia client to access the library. This work also helped to inspire a successful "Digital Library" workshop at ACM Multimedia 96 Conference (http://www.acm.org/sigmm/MM96/general.html).