The following provides a brief summary of some of the main discussion points from the fourth MODELS workshop. A fuller description is in preparation. There were 50 participants at the workshop, representing a range of communities, including libraries, museums, archives and electronic text archives, as well as systems experts. The Chair was Chris Rusbridge, Director of eLib.
The aim of the workshop was an ambitious one: to explore how resources and systems might be structured so that users can search across the UK national resource and retrieve the materials they require from a range of domains and in a variety of formats (print and electronic). The focus was on scholarly users in particular.
It was agreed at an early stage that cross-domain searching is useful, and therefore worth pursuing. However, this can be viewed at different levels: archives themselves vary so widely that cross-domain searching for archivists could mean searching across a range of archives. For others it could feasibly mean simultaneous searching across a clump of library OPACs, a museum database, a subject-based catalogue of web resources, and an electronic text archive.
The definition of 'domains' occupied some time and it was acknowledged that different interpretations should be possible. Domains could be institutions (libraries, museums, archives etc) disciplines (geology, archaeology), regions (north Wales, south-east) or others. The important point is that the availability of cross-domain searching breaks down barriers between institutions or local regions and allows access to required resources, wherever they may be held and whatever format they may be in.
The management of vocabulary control in a domain context was discussed. It is difficult, if not impossible, to envisage how vocabularies could be controlled across a range of domains. Mapping between different controlled vocabularies is more realistic.
A reiterative searching model was discussed, and emerged as desirable. At the first level is a high level generic search against simple metadata (eg Dublin Core), which will allow the user to discover an item or range of items. The second level is retrieval. At this point, in order to select the useful items, richer domain-specific metadata (eg MARC records for books, ISAD(G) records for archival material) is needed to support decisions. This is where the idea of 'drilling down' emerged: progressively moving down through a hierarchy of descriptive data to the appropriate level for the user's needs. The final stage in this model is the use of the metadata.
The meeting reached consensus that the Z39.50 protocol was potentially capable of permitting users to 'drill down' between generic and domain-specific descriptive information. However, this is a new area and work is required to explore the possibilities.
The UK archive community is currently investigating the possibilities of Z39.50 for their search and retrieval needs; a Z39.50 profile for archival materials has been drawn up and pilot testing is being carried out.
The implications for cross-domain searching of the Z39.50 Digital Collections profile 'family' were discussed. The Collection profile is an umbrella profile for navigating collections. It provides a framework for the development of extension profiles for specific domains (these are called companion profiles). Relevant companion profiles include the CIMI profile for access to museum objects, and a profile for access to digital libraries, developed initially for access to the Library of Congress digital library.
Collection descriptions are more likely to be created in some institutional contexts such as archives, than others. In fact it is not uncommon for no description to exist below collection level for archives. This is also likely to be true for collections within other domains, such as museums. Conversely, libraries tend to describe resources at item level, but not at collection level. Apart from the SCURL (Scottish Consortium of University Research Libraries) initiative using Conspectus(1), there has been no other concerted effort in the UK scholarly community to create library collection descriptions. They are however needed across all domains, in order to support clumping (see below).
Several approaches for providing collection description were identified by MODELS 3. This workshop explored more fully the possibilities offered by centroids (now becoming standardised in the Internet community under an alternative name: the Common Indexing Protocol). These are inverted indexes of database content. The workshop recommended that evaluation of the use of centroids should be carried out.
An eLib supporting study on collection description is currently being commissioned by UKOLN. The terms of reference have been extended to include museum and archival resources, in addition to library. The study is due to be completed by the end of June 1997.
Discussion of clumping arose several times during the workshop. The concept emerged during MODELS 3: the grouping together of resources to facilitate discovery. Clumps can either be fixed or dynamic. However collection descriptions are needed to support dynamic clumping.
Based on discussions and recommendations made by MODELS 3, a JISC Call for pilot virtual clump proposals appeared on 10 March. The Call is also asking for expressions of interest in cross-domain clumps, based on issues raised at MODELS 4.
With thanks to Dan Greenstein, AHDS, for his notes produced after the workshop.