This seminar was aimed at information professionals involved in the policy, development and implementation of services based on collection-level description. Such services use a single record to describe a collection as a unit, rather than recording information about its constituent parts at the item level. There has been a great deal of activity in the United Kingdom in this area since the work carried out by Michael Heaney and UKOLN in 1999 on An Analytical model of collections and their catalogues . The main focus of the seminar was to discuss the development of a standard schema for collection description with particular reference to the Research Support Libraries Programme (RSLP) schema  derived from Heaney's model, and the more recent work of the Dublin Core Collection Description Working Group  on a Dublin Core collection description application profile .
The day was split into two parts, with presentations in the morning followed by breakout sessions in the afternoon.
The seminar was chaired and introduced by Ronald Milne. He pointed out that the implementation of collection-level description had proved more complex than originally envisaged, but was still regarded as an essential building block for creating distributed national collections of print and digital resources. In the model for the Joint Information Systems Committee (JISC) Information Environment , collection-level descriptions are used to create and survey 'landscapes' which indicate potentially fruitful areas for more detailed searching at the item level. Such landscapes may be generated with a focus on subject, accessibility, geographical location, language, and other criteria for identifying concentrations of information resources. Interoperability is a key issue when descriptions from many different sources are to be used.
The first presentation was by Michael Heaney, the author of the analytical model on which most UK descriptions are based. Michael emphasised that he had taken a very abstract approach in developing the model, which was not tied to any particular implementation. The model is presented as an entity-relationship diagram which focusses on structural elements rather than transactions. The model is for a single collection and its associated finding aids, and Michael pointed out that the RSLP schema was only a partial implementation.
Ann Chapman and Bridget Robinson of UKOLN's CD Focus then gave a presentation on how the model has been implemented in a number of services in the UK. They pointed out areas where elements had been added, omitted, or relabelled to illustrate how well the model fitted service needs. They also discussed some of the generic user and manager requirements of collection-level description, and generally set the scene for the afternoon breakout sessions.
Pat Stevens of OCLC then surveyed some of the activities related to collection-level description that NISO is involved with. These include the Metasearch Initiative  and projects involving virtual reference and resource-sharing services. She emphasised the desire for a single service point for finding, identifying, selecting, and obtaining resources; for a task-based approach; for reliable indicators of collection subject strength; and for disclosure of institutional collection policies.
The last presentation of the morning was by Pete Johnston of UKOLN. Pete chairs the Dublin Core Metadata Initiative Collection Description Working Group (DCMI CD WG) , and he brought us up to date with current work in establishing Dublin Core standards for collection-level description. This involves defining vocabularies ('namespaces' in DC-speak) for labelling metadata elements and the development of an application profile which specifies how the metadata elements should be used. The profile is intended for simple descriptions suitable for a range of collections and is a partial implementation of the RSLP schema; that is, an even simpler instantiation of Heaney's model. Pete gets the prize for best diagram of the day, three concentric circles showing the relationship of DC to the RSLP schema; in other words, a CD (compact disc) view of CD (collection description). The work of Pete's group is still in progress.
For the afternoon, delegates split into three groups. One group discussed a strategy for collection-description standards while the other two concentrated on developing the standards. I was in one of the latter groups and we certainly had a lively discussion on the use of standard schemas and our practical experiences in implementing them. We clarified that the intended role of the RSLP schema (and DC CD AP) was to facilitate interoperability of descriptions from different services, rather than something that was mandatory within any specific service. The service with which I am involved, SCONE (Scottish Collections Network) , uses Heaney's full analysis rather than the cut-down RSLP schema. SCONE carried out a comparison of schemas in use in the UK in 2001  and as a result has added extra entities and relationships. SCONE descriptions map to the RSLP schema, however, and in this form can be cross-searched with records from other services. SCONE was not the only service which had used elements not included in the RSLP schema and one of our recommendations was that the schema required further development if it was to meet the needs of the UK.
This was also a conclusion of the other breakout groups. The full set of recommendations has been published . Perhaps the most significant is that future activity needs to be carried out in an international context rather than just within the UK and requires participation by appropriate national agencies.
All in all, this was a most enjoyable day despite yet another slew of acronyms and occasional mind-boggling technical complexity. As usual, it was a great opportunity to meet colleagues working in this field and catch up with the latest gossip. It was important, too, for sustaining and building on the efforts of the past five years in establishing collection-level description as a key component of the global information environment.