W3C home > Mailing lists > Public > public-rax@w3.org > July 2016

Re: Transforming RDF into XML (alternative using ShEx)

From: Felix Sasaki <fsasaki@w3.org>
Date: Sun, 10 Jul 2016 11:45:22 +0200
Cc: Jose Emilio Labra Gayo <jelabra@gmail.com>, "public-rax@w3.org" <public-rax@w3.org>
Message-Id: <B180FC90-6E9A-43A4-BB46-8AB8C5437E7E@w3.org>
To: "Evain, Jean-Pierre" <evain@ebu.ch>

> Am 09.07.2016 um 18:31 schrieb Evain, Jean-Pierre <evain@ebu.ch>:
> 
> Hi Felix, all,
> 
> I must admit I meet no difficulty going from XML to RDF.

Good for you. Note that "1) functionality needed: convert XML to RDF, or the other way round“ encompasses also: „the other way round“, and potentially also roundtripping, with or without information loss.

> I'll be happy to share files and we should have a way fo sharing such examples on github or else. I would personally by interested to actually "see" what problems others might have. Maybe I missed something :—)


I don’t think so :) a lot of this CG is about gathering what already is possible and document it. So if you can share examples and document them: please do.

- Felix

> 
> Best, Jean-Pierre
> ________________________________
> From: Felix Sasaki [fsasaki@w3.org]
> Sent: 09 July 2016 11:52
> To: Jose Emilio Labra Gayo
> Cc: public-rax@w3.org
> Subject: Re: Transforming RDF into XML (alternative using ShEx)
> 
> Thanks, Jose. From the call yesterday we came up with a rough classification of how to structure our discussion
> 
> 1) functionality needed: convert XML to RDF, or the other way round
> 2) way to implement it: with standard technologies (e.g. XSLT on the XML side) or with enhancements
> 3) enhancements being a ad hoc program (see Quentin’s approach described in a mail yesterday: "The approach is supplemented by a library of Java functions  that can be called from XSLTs to process the graph.“)
> 4) enhancements being kind of „standard“ or at least a declarative form (see in reply to Quentin: Christoph describing https://github.com/EIS-Bonn/krextor which provides an abstraction layer)
> 
> I would see ShEX falling under 4).
> 
> The reason why people do not do 2) is described by Christoph and Quentin
> 
> [
> As such, transformation between RDF and XML using XSLTs is not optimal and XSLTs need to be tailored  to different flavor of RDF/XML (which is costly).. One solution that we have used is to transform the RDF input (in any supported syntax) to be converted into a XML canonical form prior to applying XSLTs.
> 
> Indeed many manually-crafted XSLTs suffer from this,
> ]
> 
> Now, if people want to stay with 2), they can follow some potential best practice; e.g.  how to store RDF output, so that it can be processed with standard XML tools. See the mail from Martynas:
> 
> [
> We use however a canonical "flat" RDF/XML layout, in which <rdf:Description>s are not nested (default output by Jena writer). By limiting RDF/XML to such layout, and using the key() function to lookup descriptions etc., the transformation becomes quite manageable.
> ]
> 
> In the list 1-4 above *currently* the actual functionality is mixed - but we could better classify the discussion if we start with three sub tasks:
> 
> 1.1 Going from RDF > XML
> 1.2 Going from XML > RDF
> 1.3 Doing round tripping
> 1.4 Embedding RDF in XML (already feasible with technologies like RDFa or JSON-LD)
> 1.5 Embedding XML in RDF
> …
> 
> Just an idea how to structure the discussion - comments welcome.
> 
> Best,
> 
> Felix
> 
> 
> Am 09.07.2016 um 08:26 schrieb Jose Emilio Labra Gayo <jelabra@gmail.com<mailto:jelabra@gmail.com>>:
> 
> I could not attend yesterday's call because I was travelling and was not sure at what time I would arrive...
> 
> I was looking at the minutes and saw that there is interest on transforming RDF to XML. I wanted to say that one alternative could be Shape Expressions (ShEx).
> 
> Although the primary goal of ShEx is to describe and validate RDF data, it can also be extended with semantic actions which are similar to parsing grammar semantic actions and in that way, it can be used to transform the RDF that has been validated into XML or other languages like Json.
> 
> In [1] we describe ShEx and we include an example of how ShEx can be used to generate XML from RDF.
> 
> For more information about ShEx you can visit: http://shex.io<http://shex.io/>
> 
> Also, if you have interest on the subject, we have recently given a tutorial about RDF validation at ESWC. The slides are available online at: http://weso.github.io/RDFValidation_ESWC16/
> 
> [1] Shape Expressions: An RDF validation and transformation language, Eric Prud'hommeaux<http://www.w3.org/People/Eric/>, Jose Emilio Labra Gayo<http://www.di.uniovi.es/%7Elabra>, Harold Solbrig<http://www.mayo.edu/research/faculty/solbrig-harold-ralph/bio-20022700>, 10th International Conference on Semantic Systems<http://www.semantics.cc/>, Sept. 2014, Leipzig, Germany, PDF<http://labra.github.io/ShExcala/papers/semantics2014.pdf> (slides<http://www.slideshare.net/jelabra/semantics-2014>)
> 
> Best regards, Jose Labra
> 
> On Fri, Jul 8, 2016 at 3:09 PM, Felix Sasaki <fsasaki@w3.org<mailto:fsasaki@w3.org>> wrote:
> See
> https://www.w3.org/2016/07/08-rax-minutes.html
> and below as text. I created this wiki page
> 
> https://github.com/rax-w3c/draft-material/wiki
> 
> for starting editing. But I have trouble adding people to the group - can somebody help? Florent?
> 
> - Felix
> 
>   [1]W3C
> 
>      [1] http://www.w3.org/
> 
>                               - DRAFT -
> 
>                                 RAX CG
> 
> 08 Jul 2016
> 
>   See also: [2]IRC log
> 
>      [2] http://www.w3.org/2016/07/08-rax-irc
> 
> Attendees
> 
>   Present
>          felix, christian, gerard, john, timea, christoph,
>          quentin
> 
>   Regrets
>          rob, phil
> 
>   Chair
>          christian:
> 
>   Scribe
>          fsasaki
> 
> Contents
> 
>     * [3]Topics
>         1. [4]new members
>         2. [5]action items
>         3. [6]input from quentin et al.
>         4. [7]mapping XML of dictionary data to RDF
>         5. [8]other sections & topics
>         6. [9]next call in two weeks from now
>     * [10]Summary of Action Items
>     * [11]Summary of Resolutions
>     __________________________________________________________
> 
>   <scribe> chair: christian
> 
>   <scribe> scribe: fsasaki
> 
>   [12]https://lists.w3.org/Archives/Public/public-rax/2016Jul/000
>   4.html
> 
>     [12] https://lists.w3.org/Archives/Public/public-rax/2016Jul/0004.html
> 
> new members
> 
>   no new members, skipped
> 
> action items
> 
>   github link for rax cg is [13]https://github.com/rax-w3c
> 
>     [13] https://github.com/rax-w3c
> 
> input from quentin et al.
> 
>   see
>   [14]https://lists.w3.org/Archives/Public/public-rax/2016Jul/000
>   2.html
> 
>     [14] https://lists.w3.org/Archives/Public/public-rax/2016Jul/0002.html
> 
>   "One solution that we have used is to transform the RDF input
>   (in any supported syntax) to be converted into a XML canonical
>   form prior to applying XSLTs. "
> 
>   <clange> help
> 
>   processing RDF data in XML / xslt workflows
> 
>   christian: also covered by one XML prague scenario
> 
>   <quentin> canonical format new to be optimized for performance
> 
>   christoph: similar solution is here
>   [15]https://lists.w3.org/Archives/Public/public-rax/2016Jul/000
>   7.html
>   ... has RDF abstraction layer, this is independent of RDF
>   output serialization
>   ... related: what RDF abstract serialization to use
> 
>     [15] https://lists.w3.org/Archives/Public/public-rax/2016Jul/0007.html
> 
>   collection of state of the art canonical representations of RDF
>   in XML
> 
>   quentin: you should always optimize canonical format that you
>   are processsing
> 
>   <clange> there is, e.g., RXR (Regular XML RDF), there is TriX,
>   and there are more
> 
>   quentin: XSLT is generally not enough, one needs libraries on
>   top of XSLT
> 
>   christian: "processing RDF data in XML / xslt workflows" -
>   important topic, has several solutions, we need to collect info
>   what works good, what not etc.
> 
>   <clange> sources for the canonical formats mentioned above: RXR
>   ([16]https://www.dajobe.org/papers/xmleurope2004/), TriX
>   ([17]https://www.w3.org/2004/03/trix/,
>   [18]https://en.wikipedia.org/wiki/TriX_(syntax) )
> 
>     [16] https://www.dajobe.org/papers/xmleurope2004/),
>     [17] https://www.w3.org/2004/03/trix/,
>     [18] https://en.wikipedia.org/wiki/TriX_(syntax)
> 
>   <clange> AFAIK there is also a newer approach based on the
>   SPARQL XML results format
> 
>   [19]https://github.com/rax-w3c
> 
>     [19] https://github.com/rax-w3c
> 
>   <clange> @fsasaki my GitHub account is "clange"
> 
>   <TimeaT> my GitHub is theRealImy
> 
>   <quentin> @fasaki my GitHub is qhreul
> 
>   <scribe> ACTION: felix to add people to rax-cg github [recorded
>   in [20]http://www.w3.org/2016/07/08-rax-minutes.html#action01]
> 
>     [20] http://www.w3.org/2016/07/08-rax-minutes.html#action01]
> 
>   quentin: happy to do the description. won't be able to go
>   through all material that christoph provided
> 
>   christoph: happy to help
> 
>   <scribe> ACTION: quention to start section on RDF data in XML /
>   xslt workflows topic. christoph to add material [recorded in
>   [21]http://www.w3.org/2016/07/08-rax-minutes.html#action02]
> 
>     [21] http://www.w3.org/2016/07/08-rax-minutes.html#action02]
> 
>   TimeaT: previous problems, solutions were sent directly per
>   mail. there was also a template, not sure how to send input
> 
>   Christian: use the template, send to the list
> 
>   <clange> BTW my colleagues will write up their two use cases
>   (AutomationML Industry 4.0 and GATE NLP) by the next telco.
> 
> mapping XML of dictionary data to RDF
> 
>   timea: dictionary company stores data in XML format, for
>   further publishing. They wanted to bring XML into RDF, to be
>   able to enrich dictionaries
>   ... they can be multi- or monolingual, because language is
>   complex
>   ... need a way to convert XML to RDF
>   ... for working with dictionaries you have to use lemon /
>   ontolex, which may be extended
>   ... solved currently with XSLT
>   ... writing the XSLT proved to be very long and tedious
>   ... not sure if we can achieve a general XSLT
>   ... next step is how to connect words between dictionaries
>   ... we reached very good point with XSLT, just wondering if
>   there is a better way
> 
>   felix: so no integration of RDF in XML needed?
> 
>   timea: would have been desired, but dropped because XML mapping
>   took a long time
>   ... roundtripping is difficult
>   ... not sure if it is needed
> 
>   christoph: for making it easier to create XML / RDF
>   translation, there are solutions
>   ... I have developed something, felix has one
>   ... so this should be a part: write a classification of
>   existing libraries
>   ... e.g. libraries that are based on XSLT. You start with that
>   and it becomes hard to maintain, then people add an abstraction
>   layer
>   ... adding link to my implementation here
> 
>   <quentin> for RDF -> XML, an approach could be to use SPARQL
>   and XML templates
> 
>   <quentin> ... essentially creating registere queries
> 
>   <clange> [22]https://github.com/EIS-Bonn/krextor is a library
>   of XSLT templates and functions for XML to RDF
> 
>     [22] https://github.com/EIS-Bonn/krextor
> 
>   timea: we used "unified views", etl tool that read the XML and
>   writes all data into virtuoso
>   ... the actual mapping if endepend of unified views
> 
>   <clange> @Timea see
>   [23]http://ceur-ws.org/Vol-449/ShortPaper2.pdf for an old,
>   high-level overview
> 
>     [23] http://ceur-ws.org/Vol-449/ShortPaper2.pdf
> 
>   [24]http://fsasaki.github.io/stuff/feisgiltt2016/
> 
>     [24] http://fsasaki.github.io/stuff/feisgiltt2016/
> 
>   felix: currently purely XSLT based solution, without an
>   abstraction layer
> 
>   timea: classification of existing libraries, describe their
>   drawbacks etc.
> 
>   christoph: would contribute to such a section
> 
>   <quentin> depending on the complexity of the transformation,
>   you can use XPATH and templates to populate the RDF
> 
>   <scribe> ACTION: timea to start section on existing tools &
>   their classification, benefits and drawbacks - others to
>   contribute [recorded in
>   [25]http://www.w3.org/2016/07/08-rax-minutes.html#action03]
> 
>     [25] http://www.w3.org/2016/07/08-rax-minutes.html#action03]
> 
>   quentin: you can also just use XPath and java @@@ and then are
>   able to keep mappings
>   ... you can do similar things from RDF > XML perspective, using
>   SPARQL templates
> 
>   timea: that sounds familiar how the XSLT is built; has a lot of
>   templates
>   ... we transformed the XML in a simple RDF, and then converted
>   RDF into the model needed
>   ... ontolex / lemon + extension
> 
>   christioph: I have pointers to further systems, have seen a
>   related paper
>   ... will add a link here
>   ... on roundtripping, let's not forget the XSParql
>   specification
>   ... it combines XQuery and SPARQL
> 
>   <clange> SPARQL templates for RDF to XML have actually been
>   implemented as a proper language:
>   [26]https://hal.inria.fr/hal-01150623/document. XSPARQL
>   ([27]http://xsparql.deri.org/) does a similar job but it's IMHO
>   more suited for queries over XML or RDF or both at the same
>   time, rather than transforming whole RDF graphs or whole XML
>   documents.
> 
>     [26] https://hal.inria.fr/hal-01150623/document.
>     [27] http://xsparql.deri.org/)
> 
>   christioph: it is better for query than complete
>   transformation, but very interesting approaches
> 
>   <quentin> there is an overlap between the two use-case in terms
>   of transformation from 1 format to another
> 
>   felix: asking john on mark logic support for RDF and how it
>   fits into the picture
> 
> other sections & topics
> 
>   nothing at the moment
> 
> next call in two weeks from now
> 
>   same time
> 
>   john: we are working on tech. to store RDF in XML
>   ... also to make it easier to work with XML described by RDF
>   graph
>   ... interested to see in problems that people have and finding
>   solutions
> 
>   adjourned
> 
> Summary of Action Items
> 
>   [NEW] ACTION: felix to add people to rax-cg github [recorded in
>   [28]http://www.w3.org/2016/07/08-rax-minutes.html#action01]
>   [NEW] ACTION: quention to start section on RDF data in XML /
>   xslt workflows topic. christoph to add material [recorded in
>   [29]http://www.w3.org/2016/07/08-rax-minutes.html#action02]
>   [NEW] ACTION: timea to start section on existing tools & their
>   classification, benefits and drawbacks - others to contribute
>   [recorded in
>   [30]http://www.w3.org/2016/07/08-rax-minutes.html#action03]
> 
>     [28] http://www.w3.org/2016/07/08-rax-minutes.html#action01
>     [29] http://www.w3.org/2016/07/08-rax-minutes.html#action02
>     [30] http://www.w3.org/2016/07/08-rax-minutes.html#action03
> 
> Summary of Resolutions
> 
>   [End of minutes]
>     __________________________________________________________
> 
> 
>    Minutes formatted by David Booth's [31]scribe.perl version
>    1.144 ([32]CVS log)
>    $Date: 2016/07/08 12:48:48 $
>     __________________________________________________________
> 
>     [31] http://dev.w3.org/cvsweb/~checkout~/2002/scribe/scribedoc.htm
>     [32] http://dev.w3.org/cvsweb/2002/scribe/
> 
> Scribe.perl diagnostic output
> 
>   [Delete this section before finalizing the minutes.]
> This is scribe.perl Revision: 1.144  of Date: 2015/11/17 08:39:34
> Check for newer version at [33]http://dev.w3.org/cvsweb/~checkout~/2002/
> scribe/
> 
>     [33] http://dev.w3.org/cvsweb/~checkout~/2002/scribe/
> 
> Guessing input format: RRSAgent_Text_Format (score 1.00)
> 
> Succeeded: s/tempaltes/templates/
> Found Scribe: fsasaki
> Inferring ScribeNick: fsasaki
> Present: felix christian gerard john timea christoph quentin
> Regrets: rob phil
> Got date from IRC log name: 08 Jul 2016
> Guessing minutes URL: [34]http://www.w3.org/2016/07/08-rax-minutes.html
> People with action items: felix quention timea
> 
>     [34] http://www.w3.org/2016/07/08-rax-minutes.html
> 
> 
>   [End of [35]scribe.perl diagnostic output]
> 
>     [35] http://dev.w3.org/cvsweb/~checkout~/2002/scribe/scribedoc.htm
> 
> 
> 
> 
> 
> --
> -- Jose Labra
> 
> 
> ------------------------------------------------------------------------------
> 
> **************************************************
> This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed.
> If you have received this email in error, please notify the system manager. This footnote also confirms that this email message has been swept by the mailgateway
> **************************************************
> 
> 
> 
Received on Sunday, 10 July 2016 09:45:34 UTC

This archive was generated by hypermail 2.3.1 : Sunday, 10 July 2016 09:45:35 UTC