BIBFRAME Projects at the University of Washington

Joseph Kiegel

Three Projects

> Conversion and review of MARC records

> Mapping of RDA Core to BIBFRAME

> Topical collection of materials drawn from CONTENTdm, MARC and EAD, and converted to BIBFRAME

> Using LC’s BIBFRAME (bibframe.org)

Training as a First Step

> Identified a core group of librarians with format and language expertise to be trained as BIBFRAME reviewers

> Wrote in-house training with catalogers as the audience

> Used Turtle rather than RDF/XML because it is easier to read

> Opened training sessions to all interested staff in technical services

Three Training Sessions

> 1) Introduction to BIBFRAME review and training in RDF

> 2) Training in the BIBFRAME model

> 3) Detailed reading of a converted MARC record

> Later sessions were also held on RDF/XML and on Schema.org

Conversion and Review of MARC Records

Goals of BIBFRAME Review

> Familiarize ourselves with BIBFRAME

> Evaluate the BIBFRAME model

> Submit comments to the Library of Congress

Core Group of Reviewers

> Diana Brooking> Charlene Chou> Cate Gerhart> Theo Gerontakos> Joe Kiegel

> Kris Lindlan> Helice Koffler> May Rathbone> Adam Schiff> Steve Shadle

Conversion Process

> Selected and exported MARC records from the local ILS

> Converted the records to MARCXML using MarcEdit

> Converted records to BIBFRAME in RDF/XML using oXygen and the Library of Congress converter, accessed in GitHub

> Converted the RDF/XML to Turtle using a web service: rdf-translator.appspot.com

Review Sessions

> Chose records by bib format, MARC fields, or language

> Posted converted records prior to scheduled meetings

> Invited interested staff to attend

> Reviewed records in detail (1-3 records per session)

Comments for the Library of Congress

> Comments were raised as issues on the LC GitHub site

> Editorial work was centralized– To eliminate duplication of submissions– To provide some consistency in presentation

> A local issues-list was maintained so that staff could readily see what had been reported without accessing GitHub

Sample Results of the Review

> Over 125 issues were submitted

> Types of problems reported– MARC fields or subfields not converted

– BIBFRAME properties with problems

– Non-roman script not handled well

Mapping of RDA Core to BIBFRAME

Goals of the Mapping Project

> Evaluate BIBFRAME as a carrier for RDA cataloging

> Evaluate RDA/RDF as a serialization of RDA cataloging

> Focus on RDA Core as a starting point– Expand later to the full RDA element set

History of the Mapping

> Originally posted on April 17, 2015

> Updated version posted on June 9, 2015– Change Log details the revisions

> Available at:– http://faculty.washington.edu/kiegel/ld/rda-core-to-bibframe.pdf

Origin of the Document

> The RDA BIBCO Standard Record metadata application profile by the Program for Cooperative Cataloging was taken as a basis for RDA Core

> The column for MARC encoding was removed and two columns were added for RDA and BIBFRAME properties

> Some editorial changes were made to reflect RDA in a non-MARC environment

Details of the Mapping

> Some RDA elements do not map to a single BIBFRAME element

– Lightweight Abstraction Layer> rdam:titleProper maps to bf:instanceTitle >>> bf:titleValue

– Blank nodes> rdam:dateOfPublication maps to bf:publication >> bf:providerDate> rdam:dateOfDistribution maps to bf:distribution >> bf:providerDate

> Mapping may be conditional, based on the object of the RDA property

– RDA permits values to be a URI or a literal, while BIBFRAME requires the value of each property to be either a URI or a literal> rdam:mediaType http://rdaregistry.info/termList/RDAMediaType/1001> bf:mediaCategory http://rdaregistry.info/termList/RDAMediaType/1001

> rdam:mediaType “audio”> bf:mediaCategory [ a bf:Category ; bf:categoryValue “audio” ] .

> Some RDA properties do not map well to BIBFRAME

– Series treatment in BIBFRAME needs an overhaul

– Technical details are not accounted for, e.g.> Base material> Book format>Video characteristics> File type> Regional encoding

> Domains of some BIBFRAME properties do not align with the RDA WEMI model

– Domain should be unspecified in BIBFRAME, e.g.> bf:notation> bf:formatOfMusic> bf:duration

Necessity of a Carrier

> In the medium term, libraries, publishers, vendors and utilities still have separate databases

> Libraries still require a carrier to move manifestation-level descriptions, at a minimum

> With RDA as a descriptive standard, a carrier must fully support the round trip RDA BIBFRAME RDA

Examples of Round Trip Problems

> BIBFRAME does not support round trips for some elements– They map to BIBFRAME well enough, but cannot be mapped back

– Five RDA properties for extent map to bf:extent

– Four RDA properties for scale map to bf:cartographicScale

– Three RDA properties map to one for bf:note, bf:dimensions

– Two RDA properties map to one for bf:edition, bf:language, bf:originDate, bf:musicMediumNote, bf:notation, bf:relatedWork

Testing and Revision

> Created RDA cataloging manually in XML– Including a test file of all RDA core elements

> Wrote an XSLT transformation of RDA into BIBFRAME

> Evaluated the results of the transformation

> Posted a revised version of the mapping

Topical Collection of Materials in BIBFRAME

Goal of the Project

> Test BIBFRAME as a common serialization of cataloging data drawn from a variety of sources– Dublin Core– MARC– EAD

> Test searching of a corpus in BIBFRAME

Materials to Be Used

> One or two topically related collections of photographic images from CONTENTdm

> MARC records for the collections

> EAD finding aids for the collections

> Topically related materials from the library catalog in MARC

Conversion to BIBFRAME

> For CONTENTdm, export metadata in XML, and then transform it to BIBFRAME

> For MARC, export records from the ILS and convert them using the LC converter

> For EAD, write a transformation to BIBFRAME– This is likely to be the most challenging part

Evaluation of the Results

> Initial evaluation of the BIBFRAME results will be done in review sessions– Improve the mappings as a result

> After a triple store is created, we plan to load BIBFRAME data and test querying it with SQL

> Ideally, we would use a discovery interface to test the user search experience