Metadata Collections & QA


Defining library resources (an object view)

With traditional libraries, the item to be cataloged is typically an individual physical entity (object) with a beginning and an end (e.g. books, newspapers, CD-ROMs, magazines, journal articles). Secondly, many of these resources have tables of contents, indexes, or abstracts to aid catalogers. Also, users understand that if a book is a textbook about a course, they will have to search through the book for a chapter on a particular topic. In the digital library, the resource to catalog is typically not a physical entity and it is difficult to define a discrete beginning and end. There is often no table of contents, index, or abstract to aid cataloging. Additionally users would like to use a search engine to take them directly to the "chapter" of the "book" in which they are interested.

This situation makes it a challenge to determine exactly what constitutes the digital resource to be cataloged (granularity). For instance, if a 15-week web course has dynamite applets and animations that can be used in several contexts, do you catalog the course as a whole and then make separate catalog entries for the applets and animations?

To solve this dilemma, first catalog individual resources by their intended use. Then for units or graphics within a resource, catalog additional metadata records if these new records will differ substantially in educational, technical or pedagogical information. So for the above example, one would first create a metadata record to describe the 15-week course. Then catalog the individual applets and animations and relate them to the metadata record describing the course.


Defining library resources (a collections view)

With traditional libraries, a group of related objects may be thought of as a collection. Often these objects are housed together and may be built around a concept (e.g., US Civil War) or a context (e.g., a map collection). Additionally, collections can also be thought of as single objects (a Civil War collection as an object in a US history collection). In the digital world, the idea of a collection is more fuzzy due to the rapid exchanges of information and the malleable nature of digital resources.

Granularity of metadata records

  • Generally create a single metadata record for resources as a whole
  • Create additional metadata records for a whole resource if individual parts of a resource differ substantially in technical requirements, descriptions and educational information
  • Create relationships between resources using the fields of relation and occasionally, learning resource type and description (be careful using description and learning resource type because you do not want to give information that really should be cataloged in a second metadata record)
  • Apply the relationship concept above to strike a balance between supporting resource discovery with reasonable user effort and to avoid potential user frustration with retrieving too many redundant records
  • Recognize that community developed collections may have varying levels of granularity


Last updated: 1-21-04