W3C home > Mailing lists > Public > public-dwbp-comments@w3.org > December 2015

Re: About FaBiO and CiTO

From: Silvio Peroni <silvio.peroni@unibo.it>
Date: Wed, 23 Dec 2015 23:24:10 +0100
CC: <public-dwbp-comments@w3.org>
Message-ID: <4811932D-3641-4E5A-B0D9-AE330834F0F5@unibo.it>
To: Eric Stephan <ericphb@gmail.com>
Dear Eric,

> Thank you so much for your kind response and very helpful guidance on the use of SPAR ontologies.  Our citation model is very fluid at this point where it is easy to make changes and I would prefer making changes based on your recommendations prior to our request for comments in early January.

That’s great!

> We did look at biro:BibliographicRecord at one point but weren't sure about its use.  I'll look over this, if we find additional properties that are not in the scope of biro:BibliographicRecord I am wondering if our additional properties could serve as requests for additional properties within BIRO?  If we did use this I'd have to ask our other co-editors about the use of FaBIO:work.

I’m just wondering right now… I don’t know what is the right meaning of the duv:DataCitation class, but I think two alternatives are possible:

1. it is a bibliographic reference contained in an entity, e.g., a paper (i.e., the citing paper), to a particular dataset, which is then characterised by a text that, in some way, refers to the actual cited dataset;
2. it is the actual dataset that is cited by the citing entity.

If 1) is true, then it seems quite odd to me that a reference has a title and other attributes typically related with a proper bibliographic resource: they are actually attributes of the referenced work. In SPAR, that scenario could be modelled by means of the class bibo:BibliographicReference and the classes fabio:Work and fabio:Expression (or one of their subclasses). For instance, suppose that we have a paper A citing a dataset B by using the following bibliographic reference:

Doe, J. (2015). The example dataset to cite. Version 1, Sept 15, 2015. DOI: 10.3456/4567.21

Then, in SPAR, a good way for modelling this will be (I’m using Turtle now):

:dataset-b a fabio:Dataset , dcat:Dataset ;
	dcterms:title "The example dataset to cite" ;
	prism:doi "10.3456/4567.21" ;
	dcterms:creator :j-doe ;
	frbr:realization :dataset-b-v1 .

:dataset-b-v1 a fabio:Expression ;
	pav:version "Version 1" ;
	prism:publicationDate "2015-09-15"^^xsd:date .

:paper-a-work a fabio:ResearchPaper ;
	frbr:realization :paper-a-expression .

:paper-a-expression a fabio:JournalArticle ;
	frbr:part :reference-to-dataset-b .

:reference-to-dataset-b a biro:BibliographicReference, duv:DataCitation ;
	dcterms:bibliographicCitation "Doe, J. (2015). The example dataset to cite. Version 1, Sept 15, 2015. DOI: 10.3456/4567.21" ;
	biro:references :dataset-b-2015-04-01 .

:paper-a-cites-dataset-b a cito:CitationAct ;
	cito:hasCitingEntity :paper-a-expression ;
	cito:hasCitedEntity :dataset-b-v1 .

As you can see, the title, the DOI, etc., are not attributes of the bibliographic reference included in the citing paper, but are attributes of the actual cited dataset. Note that I’ve used pav:version (http://pav-ontology.github.io/pav/ <http://pav-ontology.github.io/pav/>) for indicating the version number here because dcterms:hasVersion actually relates two resources rather than a resource with a literal. Note that here the duv:hasWork has been replaced by frbr:part (in the other direction, but also frbr:partOf exist in case it is needed).

Another note: when we speak about citations, expressions (in terms of FaBiO/FRBR ontological entities) should always exist, since they actually define the actual content of a paper/dataset rather than its pure essence (i.e., the FaBiO/FRBR work). Citations, in our scenarios, should exist only when they are explicitly created by means of a text. That’s why I’ve used fabio:Expression and fabio:JournalArticle (which is a subclass of fabio:Expression) as proper entities linked to the related works (i.e., fabio:ResearchPaper and fabio:Dataset, that are subclasses of fabio:Work).

On the other hand, if the alternative 2) is true, then, we can avoid the explicit specification of such :reference-to-dataset-b, and consider our duv:DataCitation as the dataset (actually, a particular version of a dataset) cited. But then I’m not sure duv:DataCitation is actually a good name, it may be just fabio:Expression.

> It might interest you as well that we are looking at several non-semantic vocabularies expressed in XML (https://www.force11.org/sites/default/files/d7/project/882/citing-data-in-jats-2015-06.pdf <https://www.force11.org/sites/default/files/d7/project/882/citing-data-in-jats-2015-06.pdf>)

I’ve done a bit of work on JATS to SPAR conversion (see [1] for more details), and, while the mapping is always possible, it is not so direct as one can guess. I didn’t know about this new spec from Debbie, but it seems a very good XML spec to me, thanks.

> and JSON (e.g. http://okfnlabs.org/bibjson/ <http://okfnlabs.org/bibjson/>) and the BFO/IBO/IAO ontology suite .

I know them, but honestly not an expert yet.

>   I'd like to attempt map citation classes from the SPAR vocabularies to these within the scope of DUV to get buy-in about our approach in other metadata communities.  If you are aware of any work like this and can let us know, please let me know.

Beside the JATS to SPAR work, I don’t have any. However, I would be interested in working on the mapping between BibJSON and SPAR indeed…

I hope it may help.
Have a nice day :-)

S.


# Rererences
1. Peroni, S., Lapeyre, D. A., Shotton, D. (2012). Mapping JATS to RDF using the SPAR (Semantic Publishing and Referencing) Ontologies. In Proceeding of the Journal Article Tag Suite Conference 2012 (JATS-Con 2012). Bethesda, Maryland, USA: National Center for Biotechnology Information. http://www.ncbi.nlm.nih.gov/books/NBK100491/ <http://www.ncbi.nlm.nih.gov/books/NBK100491/> 



----------------------------------------------------------------------------
Silvio Peroni, Ph.D.
Department of Computer Science and Engineering
University of Bologna, Bologna (Italy)
Tel: +39 051 2094871
E-mail: silvio.peroni@unibo.it
Web: http://www.essepuntato.it
Twitter: essepuntato
Received on Wednesday, 23 December 2015 22:24:53 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 23 December 2015 22:24:54 UTC