Re: [jats-list] JATS to RDF Conversion

["Maloney, Christopher (NIH/NLM/NCBI) [C] maloneyc@xxxxxxxxxxxxxxxx" <jats-list-service@xxxxxxxxxxxxxxxxxxxxxx>]

I and some others did a little bit of work on it about two years ago now.
My impression could be summed up as bthe devil is in the detailsb. We were
trying to navigate between requiring too much purity in the semantic model
on one side (which makes the conversion very difficult) and being so loose
as to not provide good value. We had some discussions on GitHub, here:
<https://github.com/Klortho/eutils-org/issues?utf8=%E2%9C%93&q=label%3Ajats
2rdf+>, and did some work on the XSLTs, which are integrated with that
repo.

We did not get to a finished product, but if you were going to pursue it,
you might find some of the work useful.

--
Chris Maloney
NIH/NLM/NCBI (Contractor)
Building 45, 4AN36D-12
301-594-2842







"Sky Hester skyh@xxxxxxxx" wrote:

>Hi Everyone,
>
>A well-founded and clearly documented approach to the conversion of NISO
>JATS 1.0 (all tag sets)  document metadata to RDF is described in the
>article "From Markup to Linked Data" (S. Peroni, D.A. Lapeyre, and D.
>Shotton., 2012)[1]. The authors used an XSL transformation (written in
>XSLT 2.0* by Silvio) to declare the mapping *from* XML conforming to JATS
>*to* RDF XML conforming to SPAR, DCMI, and a couple of other open
>vocabularies.
>
>Many of us - at least I do - need to be able to extract semantic content
>from documents conforming to scholarly publishing standards, and I think
>the work done in the article cited above is the right place to start. Has
>anyone continued this effort since the paper? That is, do you know of any
>more recent XSLT resources for converting JATS-conformant XML to RDF
>using standard scholarly ontologies?
>
>I asked the same question to Debbie Lapeyre at Mulberry, one of the
>authors, and she explained the following:
>
>"
>	I have not [continued to work on it]. There were some complaints at the
>time that the mapping was good, but that the ontologies chosen were not
>the most commonly used, and that a mapping to more typically-used
>ontologies would have been better for the community.
>"
>
>Now that we have resources such as schema.org[2] and LOV[3], it may be
>less of a concern; for my purposes, the ontologies chosen are sufficient,
>and queries can be processed with OWL/RDFS subclasses and equivalences
>declared to preferred ones later, if needed. Then again, maybe someone
>has already produced a modified or more complete mapping using ontologies
>they prefer? I think the crucial work is in JATS coverage rather than
>names on the RDF side.
>
>Such a resource (further work on JATS2RDF) would be generally useful to
>the community.
>
>In the interest of completeness, the "Biotea"[4] paper in Journal of
>Biomedical Semantics describes an alternative RDF transformation more
>tailored to the PMC corpus than generic JATS (found on the JATS-list
>archive[5]).
>
>-Sky Hester
>
>[1] http://www.ncbi.nlm.nih.gov/books/NBK100491/
>[2] http://schema.org/
>[3] http://lov.okfn.org/dataset/lov/
>[4] http://jbiomedsem.biomedcentral.com/articles/10.1186/2041-1480-4-S1-S5
>[5]
>http://www.biglist.com/lists/lists.mulberrytech.com/jats-list/archives/201
>305/msg00028.html
>
>*I was only able to find one free implementation of XSLT 2.0, which is
>saxon-HE