MD3 - The Metadata3 Project
By jfrumkin, Section Software
Posted on Mon May 20th, 2002 at 02:36:50 PM EST
The MD3 Project takes a different approach to metadata - instead of having to create and implement new systems to handle new types of metadata, why not make new types of metadata work with our current systems?
Read on for a description of the MD3 application. Currently, the project is in its infancy (I have applied and am awaiting word on a $2000 grant to help fund development of the application), but feedback on the idea is appreciated. Update [2002-5-23 8:30:8 by jfrumkin]: I received my grant - it will provide enough money to hire a student to implement a prototype app. for the project. Yippee!
|The MD3 application provides functionality in two main areas: Metadata transformation, and Metadata sharing / retrieval. These two functions work together in a cohesive fashion to allow for a functional, scalable,and sustainable network of reusable metadata.
The metadata transformation feature of the MD3 application will allow a user to retrieve metadata in any mappable format, and transform
that metadata into any other mappable format. Of initial interest is the ability to map any metadata record (of a known format) into a
"MARC-like" record, which can then be immediately exported into a library's integrated system. The term "MARC-like" is used because MARC usually implies using AACR2 conventions, and at this point, it would be impossible to translate the actual values of a field (such as Author)
into the proper AACR2 style. However, it is very feasible to map between equivalent fields (such as DC:Creator and the MARC 100 field). The big
question here is if libraries find this level of metadata acceptable, at least as a starting point, for entry as records in their ILS's. My belief is that, given some minimal quality control, these "MARC-like" records could function quite well (as well as minimal MARC records) in an ILS, and could also be converted into "valid" MARC records much more easily and quickly then having to create the records from scratch. One can imagine a librarian finding a useful site on google, and pulling
down google's metadata record into the MD3 application, having it transformed into a "MARC-like" record, and inserted into the catalog,
complete with a working 856 field (i.e. hotlink).
The exciting part of the above is that these transformations are not limited to/from MARC records; as new metadata standards are adopted
or created, they can be incorporated into the transformation system. So, for instance, if a library ran a GIS service, and used FGDC metadata,
they could export OAI (Open Archives Initiative) compatible records that any OAI repository could harvest (or, going back to the original functionality, MARC records that any library could use).
Metadata sharing / retrieval:
Libraries have a long history of sharing metadata, namely, MARC records. Through agencies such as OCLC and RLG, libraries have had central repositories of metadata from which to browse and select. However, as more and more electronic resources become available, and the
metadata associated with these resources is presented in a variety of formats (not just MARC), libraries struggle with finding and using
appropriate metadata records which will help their customers obtain access to useful information resources. Initiatives such as CORC have
attempted to tackle this issue, but to limited success. The MD3 project is another attempt in this arena; for it to succeed, it needs to
create a single environment where users can look for metadata. The proposed mechanism for this is a "distributed repository" - this distributed repository will consist of metadata records which are stored by each MD3 application, connected via peer-to-peer (P2P) network technology. One can look at the success of applications
such as napster and gnutella to see the potential of such a network when it comes to sharing information. By utilizing a P2P approach, the need
for a managed centralized collection is eliminated; instead, the collection is distributed through the users on the network.
In addition to eliminating the need for a centralized, managed repository, the P2P approach also opens the door for a slew of
possibilities. One could imagine a mechanism for automating quality-control and link-checking via such a system (someone on the MD3 network updates a record; other MD3 applications discover this, and notify the user of the update; the user can then determine whether the update is valid, and automatically have their metadata record updated as well). There are probably many other potential features out there to be discovered. By combining metadata sharing and metadata transformation,
the MD3 application will empower libraries and librarians to utilitize metadata to its fullest extent in providing quality access to information.