Re: Re: ODS Meta Data

Gannon Dick <[email protected]> Thu, 15 Oct 2009 12:35:42 -0700 (PDT)
Newsgroups gmane.comp.openoffice.devel.xml
Message-ID <[email protected]>
My transform is 1/2 of (an unfunded) "project" to provide a compound docume=
nt to provide an archive of a data base table for Copyright purposes.  1 xm=
l archive document=3D 1 RDF section + 1 (clean) HTML table.  Normally ACCES=
S is used for this, and that has problems.  Not everyone can afford Oracle =
on their laptop (don't tell Larry E.).  In particular, I want to tie the UN=
/LOCODE system to the US system (Schedule K, Schedule D, IATA, NATO's STANA=
G etc. "Port" Codes) so you can have a memory stick sized reference. See: h=
ttp://www.rustprivacy.org/primeencoding.zip Anyway, only the meta data "sig=
nature" in RDF is important to me, for this.=0A=0AI expect our transforms a=
re alike - walk the DOM, make the leaf node attributes into elements, and r=
ecurse (does this procedure have a fancy name ?).  The only "technical" pro=
blem is the high number of triples.  About 300 is enough for a signature as=
 explained above, but if you add style and layout, etc. there are probably =
1000+ and it becomes a different problem from a Copyright perspective (and =
a forensics perspective).  Nonetheless, this is a very big deal, worldwide.=
=0A=0AI'll have a look at your tests and get back to you.  Thanks.=0A=0A--G=
annon=0A =0A=0A=0A=0A--- On Thu, 10/15/09, Svante Schubert - Sun Germany - =
Software Engineer <[email protected]> wrote:=0A=0A> From: Svante Schu=
bert - Sun Germany - Software Engineer <[email protected]>=0A> Subjec=
t: Re: [xml-dev]  Re: ODS Meta Data=0A> To: [email protected]=0A> Cc: d=
[email protected]=0A> Date: Thursday, October 15, 2009, 12:59 PM=0A> Hi=
 Gannon,=0A> =0A> I had the vision to provide one GRDDL file for creating a=
n=0A> RDF graph (file) from the ODF files (content.xml, styles.xml=0A> and =
meta.xml) for all ODF and provide it on the OASIS site=0A> =0A> =0A> Did no=
t work for months on it and would be happy if you=0A> could have a review o=
n this, before I work myself again into=0A> it and find a better on the OAS=
IS site:=0A> http://www.oasis-open.org/committees/download.php/30609/odf2rd=
f.xsl=0A> =0A> A small command line test scenario=A0 (5MB including=0A> Sax=
on.jar) can be found on the odftoolkit.org site:=0A> http://odftoolkit.org/=
projects/odfdom/downloads/download/OpenDocument-v1.2-draft%252FODF_GRDDL_TE=
ST.zip=0A> =0A> PS: After a review of the test package today, I believe I=
=0A> need to document the test coverage better, to show that=0A> a) I map a=
ll meta.xml elements and=0A> b) all cases for the content (e.g. text conten=
t from a=0A> bookmark).=0A> =0A> PPS: Must admit I did yet not fully unders=
tand your form=0A> mapping scenario. Perhaps you could you tell me what is=
=0A> missing in the GRDDL above?=0A> =0A> Thanks,=0A> Svante=0A> =0A> =0A> =
On 10/15/09 12:05, Michael Stahl wrote:=0A> > Gannon Dick wrote:=0A> ...=0A=
> >> 3. There are 20=0A> >> or so fields in a drop down box.=A0 If you want=
=0A> to use GRDDL* you'll have=0A> >> to a) take spaces out of the name and=
 b) include a=0A> namespace with the=0A> >> dot notation:=A0 e.g. "Checked =
By" would have=0A> to be typed in as=0A> >> "ODF.Checked_By", with a link t=
o "schema.ODF".=0A> > =0A> > i'm not sure about that... it depends on how y=
ou do=0A> the mapping...=0A> > if you map to one triple, then you probably =
get this=0A> problem.=0A> > if you map to two triples, one for the name and=
 one=0A> for the value, you=0A> > probably do not get this problem.=0A> > i=
n general, the XML datatype for the name is "string",=0A> so a transformati=
on=0A> > must handle all possible string values.=0A> > =0A> >> * there migh=
t be no reason for GRDDL.=A0 I've=0A> only tested it with spread=0A> >> she=
ets, but this transform should work for *.odt,=0A> etc.=A0 It's an export=
=0A> >> filter from (flat) ODF to RDF/XML.=A0 It uses=0A> the OASIS namespa=
ce as a=0A> >> default.=A0 (Warning: Don't modify it to work=0A> on the ODF=
 file, just the=0A> >> meta.xml component plus two attributes from=0A> offi=
ce:documents gives 295=0A> >> triples! see: http://www.rustprivacy.org/odf/=
sstest.zip=0A> ).=0A> >> =0A> >> The transform is :=A0 http://www.rustpriva=
cy.org/odf/ODSmeta2RDFXML.xsl=0A> > =0A> > Svante has also defined a GRDDL =
transformation, to be=0A> hosted at OASIS as=0A> > part of the ODF specific=
ation.=0A> > i hope he will answer here with more details.=0A> > =0A> >> (t=
he name is case sensitive, download it if you=0A> like and change the=0A> >=
> name)=0A> =0A> =0A> =0A=0A=0A