innovation in metadata design, implementation & best practices

DCMI Citations Working Group

Dublin Core Metadata Initiative - Citation Working Group

DC2002 Working Group Meeting, 2002-10-16 (16:30-18:30)


Chair: Ann Apps < ann.apps@man.ac.uk>, University of Manchester, UK

Main Points

  • At its October 2002 meeting the DCMI Usage Board approved as _conforming*bibliographicCitation , an element refinement of dc:identifier.
  • The Usage Board also agreed in principle to approve the Dublin Core Structured Value (DCSV) syntax, with the proviso that a DCSV must be a complete description of the property for which it is used. The Usage Board will issue an up-to-date definition of DCSV and some guidelines for its use. When these are available, the Citation WG will revise (if necessary) the DCMICite DCSV proposal to identify a journal article using its bibliographic citation information, and resubmit it to the Usage Board.
  • OpenURL does not provide a replacement for the proposed DCMICite DCSV, but could provide an alternative encoding format within a string value. The WG will produce guidelines for encoding bibliographic citation information using OpenURL. Initially this will be for the draft OpenURL version 0.1 currently in use. The guidelines will be updated for OpenURL version 1.0 when the standard is agreed and available. There are 2 problems with using OpenURL which will need to be addressed in the guidelines:
    • They are not very human readable
    • OpenURL (both versions 0.1 and 1.0) requires a base URL, or resolver.
  • It may be possible to use the XML schemas for bibliographic items being developed by OpenURL and OAI as a means of providing an XML encoding for such items within a Dublin Core record.
  • In his Keynote talk, Herbert Van de Sompel suggested that when the OpenURL registry becomes available, DC could register its scheme for bibliographic citations, then DC could be used as a 'by-value' format within an OpenURL. This is something the WG could consider after the DCMICite DCSV has been approved.
  • The WG now has a list of relevant standard identifier encoding schemes which should be registered with DCMI. This will probably be by short proposals, but these particular encoding schemes may in fact be registered by the DC-Libraries WG.
  • Citation styles. The WG will assemble a list of these, choose the most commonly used, and register them as schemes for use with dcterms:bibliographicCitation and dcterms:references.
  • To reiterate a point made 12 months ago at DC9. There is a real need for a way to encode bibliographic citation information in DC. People and projects, particularly in the library and publishing sectors, are looking for a means of capturing this information. In addition the DC Library Application Profile is dependent on the outcome of the work of this WG.

Report of DC2002 DC-Citation Working Group Meeting

1. Overview of the work of the group and the two proposals:

  • The WG has been focussing primarily on the bibliographic citation metadata for a resource itself (a citation 'in' link), not yet on linking metadata (a reference 'out' link). The WG restricted itself, in its current charter, to journal articles.
  • Journal article citation information includes identification (mostly optional and some, but not many, repeatable) of:
    • Journal: title, abbreviated title, identifier (ISSN)
    • Issue: volume, issue/part number, chronology, (identifier eg SICI)
    • Article: page range (other article information would be captured in usual DC elements) There should be sufficient information to identify the article.
  • The WG has decided that this information should be captured in dc:identifer, which recognises that the citation informationidentifies the article. This decision was informally endorsed by voting in the final Plenary session of DC8. Putting the information within a dc:relation element would not be satisfactory, because the pagination pertains to the article itself. If we were to look at linking data there is already a suitable DC property, dcterms:references (an element refinement of dc:relation).
  • During the year, the WG has submitted two proposals to the Usage Board:
    • A citation element Refinement for dc:identifier. This has now been approved by the Usage Board as conforming, but renamed bibliographicCitation.
    • A Journal Article Bibliographic Citation Dublin Core Structured Value The Usage Board has agreed in principle to approve the Dublin Core Structured Value (DCSV) syntax for encoding structured information within a text string, with the proviso that a DCSV must be a complete description of the property for which it is used. The Usage Board will issue an up-to-date definition of DCSV and some guidelines for its use. When these are available, the Citation WG will revise (if necessary) this DCMICite DCSV proposal and resubmit it to the Usage Board. It will be important to stress in this proposal that a DCMICite should contain sufficient information to identify a journal article rather than providing information about the containing journal issue.

2. Identifier Encoding Schemes. During the last year the WG has accumulated a list of relevant standard and commonly used identifier encoding schemes. This list is a working group resource and should be maintained. Some of these schemes should be registered with DCMI. This will probably be by short proposals, but these particular encoding schemes may in fact be registered by the DC-Libraries WG. Another possibility would be to encode identifiers as URIs, eg. urn:issn:0740-8188.

3. OpenURL. An overview was given of the development of the NISO OpenURL standard. It has been proposed as a means of encoding bibliographic citation information in a machine-parsable way within the string value of a DC property. It is DCMI policy to utilise other standards where possible. OpenURL covers more material types than just journal articles, including books and conference papers, so would save the WG from having to define these. The WG thinks that OpenURL does not provide a replacement for the proposed DCMICite DCSV, but could provide an alternative encoding format within a string value. The WG will produce guidelines for encoding bibliographic citation information using OpenURL 'by-value' metadata, because people will want to encode bibliographic citation in this way. Possibly this will be part of a wider 'Guidelines for Encoding Bibliographic Citation Information in DC'. Initially this will be for the draft OpenURL version 0.1 currently in use. The guidelines will be updated with OpenURL version 1.0 when the standard is agreed and available. There are 2 problems with using OpenURL which will need to be addressed in the guidelines:

  • They are not very human readable. This could be overcome by suggesting that best practice would involve having an additional human-readable citation, or always presenting the information to the end-user through an application which presents the information in a user-friendly way.
  • OpenURL (both versions 0.1 and 1.0) requires a base URL, or resolver. Previous suggestions for using OpenURL have in fact been a 'partial OpenURL' without the resolver, which is not a standard OpenURL. The solution to this problem would seem to be:
    • Either, accept a 'not quite standard' use of OpenURL
    • Or, require a dummy resolver base URL to be included

In his Keynote talk, Herbert Van de Sompel suggested that when the OpenURL registry becomes available, DC could register its scheme for bibliographic citations, then DC could be used as a 'by-value' format within an OpenURL. This is something the WG could consider after the DCMICite DCSV has been approved. It seems sad that the development of a means of encoding journal article bibliographic citation information in DC has taken so long. If it had already been in place, it is likely that OpenURL would have adopted it.

4. XML Encoding. The proposed DCMICite DCSV is just for a structured string representation of citation information. Some applications will want to encode this information in XML. It may be possible to use the XML schemas for bibliographic items being developed by OpenURL and OAI as a means of providing an XML encoding for such items within a Dublin Core record. This is something the WG should work on, following the Encoding DC in XML Guidelines. Some input may be needed from the DC-Architecture WG.

5. Citation Styles. It has been suggested that some Citation styles could be used as encoding schemes with dcterms:bibliographicCitation or dcterms:references. The WG has already begun to assemble a list of these which will continue. The WG will then choose a few, which are most commonly used, to support. The most likely are APA and Chicago. There is no intention to include every publisher's house style. They will be registered as encoding schemes for use with dcterms:bibliographicCitation and dcterms:references. Some reasons for doing this work:

  • Librarians would like to note the style when they create a citation in metadata
  • Because they are structured they would assist an application in user presentation of a citation. They tend to be community specific so would allow for customisation of display for each community

6. Open Issues.

  • Citations for linking. This could be taken care of by using the currently proposed encodings, in particular OpenURL, within dcterms:references.
  • Other material types. We may decide to extend the DCMICite DCSV for books, but probably other types could be encoded using OpenURL
  • Electronic only journal articles. These do not have page numbers. On the other hand they should have URLs, so may not need to be considered separately
  • Author affiliation for journal articles. The affiliation of a scholarly work persists even when an author moves. This could be solved by including the affiliation institution as a 'contributor'. The 'agents linking' work currently being done by the DC-Agents WG may provide a solution
  • A 'journal article' type. This should be covered by some standard typelist such as that from the DC Library Application Profile

2002-11-07

[Dublin Core Used Here]