Lists Project – making collection lists searchable through Trove.

15
Lists Project – making collection lists searchable through Trove

Transcript of Lists Project – making collection lists searchable through Trove.

Lists Project – making collection lists searchable

through Trove

List Project Report• The Lists Project set out to investigate

the issues involved in making lists that describe collections searchable online.

• The work included creating models of ways to convert the lists into metadata and how to chose one of those methods.

Bibliographic data models summary

• Model 1 – List as digital image attached to collection level MARC record

• Model 2 – List as indexed data searchable via Trove

• Model 3 – List as enhanced collection level MARC record

• Model 4 – List as item level MARC records

Model One – List as digital image attached to collection level MARC record.

The List - Farsi collection

MARC record showing link to list.http://catalogue.nla.gov.au/Record/4963622

Model Two – List as machine readable text indexed by Trove.

The List - Griffin and his work

Terms from the Griffin list searched on Trove.http://trove.nla.gov.au/work/32799480

MARC record showing link and indexing flag

Model Three – List is converted to collection level MARC record, enhanced by list terms.

The List - Finding aid for Giblin papers

Papers of Giblin MARC record enhanced with list terms. http://catalogue.nla.gov.au/Record/2950469

Model Four – List converted to item level MARC records.

The List – Baulkham Hills plans spreadsheet.

One of the Baulkham Hills item level records.

Nature of lists

• The nature of the list and the collection it represents will influence the metadata structure and choice of data model.

• The characteristics of the data on the list will affect the method and required resources to convert the list to metadata.

Conclusions about lists

• Machine readable text is paramount to most conversions.

• Homogeneity of data makes conversion easier.

• Finer granularity of access can only be usefully provided if distinctive data is available.

• Item format and published status will determine match/merge loading profiles.

More conclusions about lists

• A model choice can be unfavourable if it relies on good MARC record linking which is unavailable in some systems.

• Data format and arrangement determines the ease and expense of list conversions.

• To review or not to review the data for quality and currency is a difficult decision.

• If the list is hierarchical in nature like archival finding aids, there will need to be compromise between collection context and metadata presentation.