CollectionLevel Description Gordon Dunsire Depute Director Centre for

Collection-Level Description Gordon Dunsire Depute Director, Centre for Digital Library Research Presentation for a workshop at the Libraries in the Digital Age Conference, May 26 -30 2003, Dubrovnik

Overview • • What is a collection? What is collection-level description? Why is it important? Development in UK Some practical issues Scottish Collections Network (SCONE) Information environments

What is a collection? • “Any aggregation of individual items (objects, resources)” – – CD Focus briefing paper 1 Size is not a factor – 1 item is possible Varying degrees of permanence Physical juxtaposition not necessary; collections can be distributed across multiple locations • Cross-domain – Libraries, museums, art galleries, archives, digital • Definition is too vague to be practicable – Limit to “useful” collections • “Useful” defined in terms of “Functional granularity”

Functional granularity • “… useful or necessary for the purposes of resources discovery or collection management” – Heaney – As deemed by “the institution” – Might include user groups as well as owners and administrators • Exclude – Dynamic collections (results of retrieval) – Single persons (unless significant)

What is CLD? • Collection-Level Description – Metadata at the level of aggregation: Title: William Speirs Bruce Collection Description: Collection of material on oceanography and Arctic and Antarctic exploration, bequeathed by Dr. William Speirs Bruce, Polar explorer and oceanographer (1867 -1921). Location: Edinburgh University Library. Main Library Collectors: William S. (William Speirs) Bruce (1867 -1921) [Collecting: Closed] Subjects: Antarctica--Discovery and exploration Part of: Edinburgh University Library. Department of Special Collections printed books collections

Confusing terms • Collection-Level Description – The complete metadata for a collection – The process of creating a CLD • Collection-Description – A finding-aid for the collection (e. g. catalogue) • Description – An attribute of a Collection giving a short summary of the collection history and contents, etc.

Why is CLD important? (1) • Ideally, all metadata/retrieval is at the level of the work (item-level description) • But in the Real world … – Online ILD metadata not available • Legacy; Institutional policies – Wide variation in ILD structure and content standards • Between domains; within domains • Within single institutions!

Why is CLD important? (2) • CLD offers broader coverage – More stuff can be found – Cheaper to implement – High recall, low precision • Some metadata cannot be accommodated in ILD without extensive duplication – E. g. Collection title, Collector, Owner, Location, etc.

Why is CLD important? (3) • Collaborative management – Collaborative acquisition policies – Preservation and storage – Priorities for digitisation, wider access, etc. • Landscaping in distributed digital information environments – Portals – Broad overview, then more precise discovery

Landscaping Search term or Profile parameter e. g. name, subject, education level, accessibility Retrieve relevant CLDs to create broad "map" of concentrations of resources: peaks of significance; "lodes" for further exploration CLDs link to digital collections, and online (analytic) finding aids Local ILDs for resource discovery: cross-searching possible with Z 39. 50/OAI

Development of CLD in UK • Entity-relationship model – Michael Heaney – Also covers analytic finding aids: collectiondescriptions (C-Ds) • Database schema – For RSLP by UKOLN; simplifies Heaney’s model • Implementation – JISC IE Services Registry; simplifies RSLP

Heaney’s Analytic Model

Heaney's components • Entities – Collection; Agent; Location • Relationships – Collection: Is. Located. In: Location – Administrator[Agent]: Administers: Location – Collector[Agent]: Collects: Collection – * Collection: Has. Part: Collection – * Collection: Is. Described. By: C-D[Collection] * Heaney focussed on single collections

CLD in practice (1) • Collection titles – If no specific title, derive from name of institution or user group defining the collection • Collection hierarchies – Multi-level granularity (6 levels in SWOP) – Polyhierarchy: one physical super-collection, but many virtual – Data redundancy; inheritance from supercollection • E. g. location, owner, access

CLD in practice (2) • Content interoperability – Cross-searching names and subjects in landscapes – Varying standards in different organizations • Agent names (persons and organizations) – Much wider range than item-level description • Owners, administrators in addition to creators, subjects, to be included in name authority files • Subjects – Collections on specific subjects – General collections; subject strengths

CLD in practice (3) • Dates – 18 th century books on classical Greece collected from 1890 to 1930 – dates of: manufacture; subject; aggregation • Significance – Quantity vs quality; subjective; dynamic – 5 first editions with manuscript notes by Robert Burns, or 50000 items by and about Burns?

SCONE story (1) • CAIRNS – Z 39. 50 clump for distributed searching – Metadata for Z servers (service-level description!) – Associated metadata for collectiondescriptions (catalogue indexes, etc. ) – Associated metadata for CLDs – Access (SQL) database

SCONE story (2) • SCONE project – Collaborative collection management • HE/FE plus public libraries sector (SEED) • CDLR as lead site – Test datasets • SLIR; SWOP; ESH; Websites – Then Heaney's model and RSLP schema • SCONE service – 2600 CLDs

SCONE story (3) • SQL database (MS SQL Server) – Uses Heaney’s analysis rather than RSLP – Fully relational, normal form – Incorporates additional metadata not specified • Subject strengths (RCO) • Service-level description elements (CAIRNS) • Cold. Fusion Web data server • Dream. Weaver Website maintenance

SCONE futures • CC-interop (COPAC/Clumps interoperability) project – Cross-relates SCONE to major UK schemas – SCONE clone for RIDING clump • Ha. IRST (institutional resources) and SPEIR (Scottish portals) projects – SCONE used for landscaping

JISC Information Environment • “the set of network or online services that support publishing and use of information and learning resources” • Functional model for resource discovery has 4 stages – Observes that some components already exist or are under development

JISC IE Functional Model • 1: Enter – Initial landscape: presentation of collections & services for local service or user profile • 2: Survey – Modify set of collections & services • 3: Discover – Item-level searching using distributed (z 39. 50) or physical (OAI harvested; FTP) union catalogue • 4: Detail – Further information about items

IE for Scotland (A) Entry Initial landscape [Scottish Cultural Portal; SCONE] Survey Collection descriptions service [SCONE] Landscaper Collection-level descriptions

IE for Scotland (B) Discover Distributed union catalogue [CAIRNS] Harvested union catalogue [Ha. IRST] Union catalogue [COPAC] Detail Item metadata

Links • Me – g. dunsire@strath. ac. uk • SCONE service – http: //scone. strath. ac. uk/service/index. cfm – “About SCONE” for more information • CDLR (other projects) – http: //cdlr. strath. ac. uk • JISC Information Environment – http: //www. jisc. ac. uk/index. cfm? name=about_info_en v