The Metadata Switch is an umbrella activity for a set of projects which are constructing experimental modular services that add value to metadata. A partial listing of such services includes:
- harvesting metadata
- 'fusion' of metadata from different sources
- schema transformation
- enrichment or augmentation of records with various types of data
- terminology and name authority services
We will also be testing how to expose this functionality as 'web services', modular web-based machine-to-machine applications which can be combined in various ways.
Metadata Switch projects include:
- Schema Transformation Services
- Recombinant Catalog Metadata
- Terminology Services
|Examples of potential Web-based Metadata Switch services|
|A service that accepts:||...and then returns:|
|1. a record in one schema (say, Dublin Core)||the record in another schema (say, MARC)|
|2. a document or web page||a classification number or subject heading|
|3. a class number||an indication of whether it is valid or not|
|4. a name||a list of candidate name matches from an authority file|
|5. a name||a cluster of names which appear to be the same person, pulled from other databases|
|6. a record||the record enriched with data from other records for the same item|
|7. a class number or subject heading from one schema||a mapping in another schema|
When we look at the 'metadata landscape' we see several new directions.
- Libraries are creating more metadata for more types of material.
- They are using different metadata formats
- In some cases they may use different metadata systems.
So for example, it would not be unusual
- for a library to be creating metadata in some or all of
- Dublin Core
- Encoded Archival Description (EAD)
- or for this activity to take place within
- a library management system
- a content management system
- a custom-developed database application.
We are also seeing a growing interest in harvesting metadata, pulling metadata from different repositories, fusing it and turning it around in 'union' services. The mechanics of harvesting are becoming routine and well understood, and this is now introducing the interesting challenge of effectively fusing metadata so that a useful retrieval experience can be offered.
This metadata will often not have been created within a framework of consistent practice; approaches to subjects or names will be different for example. The question we have asked ourselves is what type of services would be valuable to libraries in this increasingly diverse environment.
The Metadata Switch project is a response to this question. We want to try to make sure that the accumulated expertise and value invested in libraries' metadata management and knowledge organization approaches is deployable across the range of need.
Some of the factors influencing this approach are:
- Much of the value of what we offer is embedded in integrated systems which meet a central need, but which may not support the changing working patterns of all our potential users.
- We need to explore how to break down services into components which our users can build back up within their environments and workflows (example: a service which took converted terms from one subject scheme to another could be used in a metadata creation environment; it could also be used by a search system in a multi-database environment).
- We also need to figure out how to make our existing investment in structured metadata work more for us by mining, developing, and exposing relationships across documents and other resources.
In parallel with technical development, work will be done on testing and developing market demand, and on potential business models, for such services, beginning with an invitational conference of expert practitioners.
To learn more about Web services, see:
- Roy Tennant. Digital Libraries-What To Know About Web Services. Library Journal (July 15, 2002).
- Tracy Gardner. An Introduction to Web Services. 2001 Ariadne 29 (October 2001).