RE: Feedback on Ingredients for High Quality Linked Data section of Linked Data Cookbook

Dear Charles,

> If you
> know someone who might be able to provide some guidance as to what the
> RDF/OWL would look like (I want to provide a good example), I would be
> happy to take their guidance in creating such a prototype.
We have tried to create such a prototype with the Edgar Linked Data Wrapper
[1]. It uses the Google App Engine, Java, and XSLT-Sheets to make available
XBRL filings as Linked Data.

As an example, take an XBRL filing of RAYONIER INC disclosing a sales
revenue net of 377,515,000 USD from 2010-07-01 to 2010-09-30 [2].

As Linked Data published by Edgar Linked Data Wrapper, this filing is
represented by a URI: [3].

If you open this URL with a browser you get the RDF representation of that
filing. 

The RDF representation reuses the RDF Data Cube Vocabulary (QB). Modeled
using QB, a filing is basically a dataset with facts.

For instance, "RAYONIER INC disclosing a sales revenue net of 377,515,000
USD from 2010-07-01 to 2010-09-30" is represented as follows in RDF. Note
the different dimensions of the fact, subject (financial concept), dtstart,
dtend (period), issuer (filing company), and the measure (observation
value).

<rdf:Description rdf:about="#ds">
<rdfs:seeAlso>
<qb:Observation>
	<qb:dataSet rdf:resource="#ds"/>
	<ed:subject
rdf:resource="http://edgarwrap.ontologycentral.com/vocab/us-gaap-2009-01-31#
SalesRevenueNet"/>
	<sdmx-measure:obsValue>377515000</sdmx-measure:obsValue>
	<ical:dtstart>2010-07-01</ical:dtstart>
	<ical:dtend>2010-09-30</ical:dtend>
	<ed:issuer rdf:resource="../../cik/52827#id"/>
</qb:Observation></rdfs:seeAlso></rdf:Description>

This example is the best recommendation I can give you, at the moment.
Certainly, the model can be improved; for instance, it does not make
explicit typed and explicit axes and attributes such as the conversion unit;
it also does not reflect taxonomies or linkbases, yet. But we were able to
use this model in our FIOS demo [4]. The model is a start which - if time
permits - we would be happy to improve upon.

Last week at the weekly telecon of the GLD Working Group, I gave a
presentation about making XBRL and Linked Data interoperable. We were
discussing about having XBRL as a use case for our work on the RDF Data Cube
Vocabulary. So far, no decision has been made. If you are interested, see
[5] for the presentation.

Best,

Benedikt

[1] <http://edgarwrap.ontologycentral.com/> 
[2]
<http://www.sec.gov/Archives/edgar/data/52827/000119312510238973/0001193125-
10-238973-index.htm> 
[3]
<http://edgarwrap.ontologycentral.com/archive/52827/0001193125-10-238973#ds>

[4] <http://xbrl.us/research/appdev/pages/275.aspx#> 
[5] <http://www.aifb.kit.edu/images/c/c3/Kaempgen_QB-XBRL_2012-05-17.pdf> 

-- 
AIFB, Karlsruhe Institute of Technology (KIT)
Phone: +49 721 608-47946 
Email: benedikt.kaempgen@kit.edu
Web: http://www.aifb.kit.edu/web/Hauptseite/en 



> -----Original Message-----
> From: Charles Hoffman [mailto:CharlesHoffman@olywa.net]
> Sent: Wednesday, May 16, 2012 4:47 PM
> To: 'Benedikt Kämpgen'; 'Frankel, David'; public-gld-comments@w3.org
> Cc: 'O'Riain, Sean'; 'Andreas Harth'; 'Herm Fischer'
> Subject: RE: Feedback on Ingredients for High Quality Linked Data section
of
> Linked Data Cookbook
> 
> Benedikt;
> 
> Seems like we are on the same page.
> 
> I agree with you that the RDF/OWL I showed you does not reflect an entire
> SEC financial filing.  I am moving slowly in that direction only because I
don’t
> know RDF/OWL.  I am working with someone now who understands
> modeling, but they likewise are not that experience with RDF/OWL.  If you
> know someone who might be able to provide some guidance as to what the
> RDF/OWL would look like (I want to provide a good example), I would be
> happy to take their guidance in creating such a prototype.
> 
> When I get the RDF/OWL model correctly reconciled to my model, I can (and
> will) get the entire set of SEC XBRL financial reports generated in that
format.
> It will just be a matter of creating a few style sheets which converts my
> existing model to the RDF/OWL output format.  That is exactly what I am
> doing now to get to my model.  I am using someone else's model, using
style
> sheets to convert it to my model.
> 
> The model is not the problem, I am 100% confident that (a) I understand
the
> financial report model and (b) that model works with 100% of SEC XBRL
> financial filings.  My model was actually created using many, many inputs
(i.e.
> it was not me who came up with the model, I just put all the pieces
> together).
> 
> I think this would be a great exercise and would provide a boat load of
useful
> RDF/OWL for people to play with, particularly starting in July when 100%
of
> SEC filers are providing detail in XBRL.
> 
> I figure that converting my models to RDF/OWL would take less than a day
if
> the person doing it understands the goal is to simply convert my model;
not
> discuss what the model should be.  If desired, we could then adjust my
> model as a second step, correcting any errors.
> 
> Cheers,
> 
> Charlie
> 
> -----Original Message-----
> From: Benedikt Kämpgen [mailto:kaempgen@fzi.de]
> Sent: Wednesday, May 16, 2012 7:16 AM
> To: 'Charles Hoffman'; 'Benedikt Kämpgen'; 'Frankel, David'; public-gld-
> comments@w3.org
> Cc: 'O'Riain, Sean'; 'Andreas Harth'; 'Herm Fischer'
> Subject: RE: Feedback on Ingredients for High Quality Linked Data section
of
> Linked Data Cookbook
> 
> Dear Charles,
> 
> > That RDF stuff you sent me (http://edgarwrap.ontologycentral.com/) is
> > awesome!!!
> Great. We can give you more information about this prototype, if you want.
> 
> > With all due respect, I believe you are misinterpreting what I am
> > sending you or something.  The infosets I sent to you have 100% of the
> > properties you need to interpret the data 100% correctly.  Actually,
> > that is not totally true, I did not provide you with the business
> > rules
> infoset.
> > Whether my infoset is best or has the right information is really not
> > the correct discussion as I see it.  This was the discussion I was
> > trying to have, I have condensed it into bullet points.
> I think, I have quite a clear picture of what you sent around; I also
think, we
> have the same opinions. Only, let me maybe put my point differently:
> 
> You linked to two XML files:
> 
> * Ontology for Digital Financial Report:
> http://www.xbrlsite.com/DigitalFinancialReporting/Ontologies/2012-04-
> 15/Digi
> talFinancialReport.xml
> * Ontology for Multidimensional Model:
> http://www.xbrlsite.com/DigitalFinancialReporting/Ontologies/Multidimensi
> ona
> lModel.xml
> 
> I think they are a great start, but I am wondering how these two XML files
> can be used to represent a concrete XBRL filing or XBRL taxonomy. I
haven't
> found any examples.
> 
> > This thread started because I pointed out that the Government Linked
> > Data Working Group "Ingredients for High Quality Linked Data"
> >
> (http://www.w3.org/2011/gld/wiki/Linked_Data_Cookbook#Ingredients_for
> > _High_Q
> > uality_Linked_Data) had nothing in the list which says "verify that
> > the
> data
> > you are making available as linked data is correct and make those
> > business rules available."
> Understood. The paper on "Using SPIN to Formalise Accounting Regulations
> on the Semantic Web" I mentioned presents a way to represent (and even
> to
> execute) those business rules for consistency checks on financial reports
in
> RDF.  Yet, you are right, business rules is something we could expand in
the
> Linked Data Cookbook.
> 
> As far as the the RDF Data Cube vocabulary (QB) standard alone is
concerned,
> business rules are probably out-of-scope, since QB aims at a generally
> applicable way to represent statistics (such as contained in financial
reports),
> and as such focuses more on interoperability and exchange, than on
> consistency checks and automatic inferencing of additional statistics.
> Still, representing XBRL information as Linked Data may be a nice use case
for
> QB.
> 
> Thanks for summarising our discussion.
> 
> Best,
> 
> Benedikt
> 
> 
> > -----Original Message-----
> > From: Charles Hoffman [mailto:CharlesHoffman@olywa.net]
> > Sent: Friday, May 11, 2012 4:36 PM
> > To: 'Benedikt Kämpgen'; 'Frankel, David'; public-gld-comments@w3.org
> > Cc: 'O'Riain, Sean'; 'Andreas Harth'; 'Herm Fischer'
> > Subject: RE: Feedback on Ingredients for High Quality Linked Data
> > section
> of
> > Linked Data Cookbook
> >
> > Benedikt;
> >
> > That RDF stuff you sent me (http://edgarwrap.ontologycentral.com/) is
> > awesome!!!
> >
> > With all due respect, I believe you are misinterpreting what I am
> > sending you or something.  The infosets I sent to you have 100% of the
> > properties you need to interpret the data 100% correctly.  Actually,
> > that is not totally true, I did not provide you with the business
> > rules
> infoset.
> >
> > Whether my infoset is best or has the right information is really not
> > the correct discussion as I see it.  This was the discussion I was
> > trying to have, I have condensed it into bullet points.
> >
> > Respectfully, I would hold the following out as facts:
> >
> > 1. Some people use XBRL (i.e. the SEC), some people use RDF (i.e. your
> > RDF demo), some people use proprietary XML (my infoset) to express
> > financial information.  This is syntax.
> >
> > 2. XBRL, RDF, and other expressions of the same financial information
> should
> > mean EXACTLY the same thing semantically.
> >
> > 3. XBRL has a business rules engine built it which can be used to
> > verify information expressed in the XBRL technical syntax.
> > Proprietary XML does not and RDF does not have that capability either.
> > However, you could express the business rule information in XML or RDF.
> >
> > 4. If RDF and any proprietary XML has the same semantics and all the
> > appropriate properties are expressed, one can convert RDF to XBRL and
> > use an XBRL processor to verify information; or convert any
> > proprietary XML
> format
> > into XBRL and use an XBRL processor to verify the information.
> >
> > 5. This ability to convert from the RDF technical syntax or any other
> > proprietary XML technical syntax to/from XBRL is both useful and
> desirable.
> >
> > 6. The ONLY think necessary to achieve number "5" above is an
> > understanding of the semantics.
> >
> > THAT is why XBRL International needs to document those semantics. That
> > is why I got David Frankel on this thread, that is what he is trying
> > to
> do.
> >
> > Personally, I believe that RDF is more important than XBRL because RDF
> > is
> a
> > W3C standard.  The fact is when XBRL first started, they considered
> > using RDF to express what has been expressed using XBRL but RDF was
> > not mature enough at that time.
> >
> > This thread started because I pointed out that the Government Linked
> > Data Working Group "Ingredients for High Quality Linked Data"
> >
> (http://www.w3.org/2011/gld/wiki/Linked_Data_Cookbook#Ingredients_for
> > _High_Q
> > uality_Linked_Data) had nothing in the list which says "verify that
> > the
> data
> > you are making available as linked data is correct and make those
> > business rules available."
> >
> > I think that would be a good addition to your ingredients and it would
> help
> > XBRL and RDF be interoperable.  That is an ingredient which I use for
> > my XBRL-based information and those infosets (which is just an easier
> > to use form of exactly the same information expressed in XBRL).
> >
> > I say this as a CPA and accountant who understands the importance of
> > making sure that things "tick and tie", "cross cast and foot".  That
> > is a
> business
> > requirement, a business use case in general and it is absolutely a
> business
> > requirement for financial information.  I pointed out the example of
> > the
> US
> > Census Bureau making information available which was incorrect, and
> > they did not even know that it was incorrect.  I found that it was
> > incorrect by checking my instantiation of that data in XBRL.
> >
> > So, that is what I am trying to say.  Which, if any, of those points
> > would you disagree with?
> >
> > Cheers,
> >
> > Charlie
> >
> > -----Original Message-----
> > From: Benedikt Kämpgen [mailto:kaempgen@fzi.de]
> > Sent: Thursday, May 10, 2012 3:20 PM
> > To: 'Charles Hoffman'; 'Benedikt Kämpgen'; 'Frankel, David';
> > public-gld-comments@w3.org
> > Cc: 'O'Riain, Sean'; 'Andreas Harth'; 'Herm Fischer'
> > Subject: RE: Feedback on Ingredients for High Quality Linked Data
> > section
> of
> > Linked Data Cookbook
> >
> > Dear Charlie,
> >
> > Thank you for this discussion.
> >
> > I was not referring to your semantic model/reference implementation. I
> > meant the ontology at [1] which describes well classes of a
> > multidimensional
> model
> > but does not define object properties between instances of such classes.
> > This would render representation of XBRL filings or taxonomies reusing
> > the ontology difficult. For instance, how could one describe that a
> > dimension
> is
> > part of a specific hypercube or that a fact has a specific dimension
> > and member?
> >
> > The RDF Data Cube Vocabulary (QB) includes such properties to fully
> > represent a multidimensional model. I would be happy to explain the
> > correspondence to you in more detail if you like. Also, we have
> > published
> a
> > paper on mapping QB and a multidimensional model [2].
> >
> > Thanks for the pointer to your new ontology [3], which is more
> > specific to financial reporting than the previous ontology and which
> > introduces
> classes
> > of the Financial Report Semantics and Dynamics Theory. However,
> > describing instances of filings (or even taxonomies) would again be
> > difficult, since the ontology is missing object properties.
> >
> > Maybe you are interested in how we are currently describing XBRL
> > filings using QB. The Edgar Linked Data Wrapper [4] translates filings
> > from the
> SEC
> > on-the-fly into Linked Data. See [5] for an example filing. Besides
> > QB, we are reusing other ontologies, e.g., RDFS (just like you), and
> > SKOS [7]. If you are wondering about the difference between Linked
> > Data and OWL, Linked Data (see [6] for a description by Tim
> > Berners-Lee) refers to general best practices of making RDF available
> > on the Web, whereas OWL is one specific ontology language that can
> > also be represented in RDF.
> >
> > Already quite a few ontologies/vocabularies to describe XBRL
> > filings/taxonomies can be found in the literature, but maybe none of
> > them does it properly. It would be great, however, if your ontology
> > would be linkable to QB (or even better, reuse QB). This way, every
> > application
> that
> > works with statistics described with QB, would also work with filings
> > published using your ontology.
> >
> > I hope what I meant is clearer now.
> >
> > > Yes, there is validation software. This is an example of validation
> > > which
> > has
> > > been run against every SEC XBRL financial filing:
> > > https://edgardashboard.xbrlcloud.com/edgar-dashboard/dashboard.do
> > Thanks for this interesting pointer.
> >
> > Best,
> >
> > Benedikt
> >
> > [1]
> > <http://www.xbrlsite.com/DigitalFinancialReporting/Ontologies/Multidim
> > en
> > sion
> > alModel.xml>
> > [2] <http://www.aifb.kit.edu/web/Inproceedings3211/en>
> > [3]
> > <http://www.xbrlsite.com/DigitalFinancialReporting/Ontologies/2012-04-
> > 15/Dig
> > italFinancialReport.xml>
> > [4] <http://edgarwrap.ontologycentral.com/>
> > [5] <edgarwrap.ontologycentral.com/archive/909832/0001193125-10-
> > 230379#ds>
> > [6] <http://www.w3.org/DesignIssues/LinkedData.html>
> > [7] <http://www.w3.org/TR/skos-reference/>
> >
> > --
> > AIFB, Karlsruhe Institute of Technology (KIT)
> > Phone: +49 721 608-47946
> > Email: benedikt.kaempgen@kit.edu
> > Web: http://www.aifb.kit.edu/web/Hauptseite/en
> >
> >
> >
> > > -----Original Message-----
> > > From: Charles Hoffman [mailto:CharlesHoffman@olywa.net]
> > > Sent: Thursday, May 10, 2012 7:27 PM
> > > To: 'Benedikt Kämpgen'; 'Frankel, David'; public-gld-comments@w3.org
> > > Cc: 'O'Riain, Sean'; 'Andreas Harth'; 'Herm Fischer'
> > > Subject: RE: Feedback on Ingredients for High Quality Linked Data
> > > section
> > of
> > > Linked Data Cookbook
> > >
> > > See this blog post of mine which explains the best attempt I can
> > > make at creating an RDF/OWL representation of what is in the other
> > > documentation I pointed you to.
> > >
> > > This is the blog post (it has a graphic generated via Protégé):
> > >
> > > http://xbrl.squarespace.com/journal/2012/5/10/digital-financial-repo
> > > rt
> > > -
> > > seman
> > > tics-expressed-using-rdfowl.html
> > >
> > > This is the RDF/OWL ontology:
> > >
> > > http://www.xbrlsite.com/DigitalFinancialReporting/Ontologies/2012-04
> > > -
> > > 15/Digi
> > > talFinancialReport.xml
> > >
> > > This has a ways to go, but hopefully within two or three months I
> > > will
> > have
> > > both the ontology build out more and I will have converted every SEC
> > > XBRL financial filing into RDF following this ontology.
> > >
> > > Cheers,
> > >
> > > Charlie
> > >
> > >
> > >
> > > -----Original Message-----
> > > From: Charles Hoffman [mailto:CharlesHoffman@olywa.net]
> > > Sent: Monday, May 07, 2012 8:46 AM
> > > To: 'Benedikt Kämpgen'; 'Frankel, David'; 'public-gld-comments@w3.org'
> > > Cc: 'O'Riain, Sean'; 'Andreas Harth'; 'Herm Fischer'
> > > Subject: RE: Feedback on Ingredients for High Quality Linked Data
> > > section
> > of
> > > Linked Data Cookbook
> > >
> > > Benedikt;
> > >
> > > Yes, there is validation software. This is an example of validation
> > > which
> > has
> > > been run against every SEC XBRL financial filing:
> > >
> > > https://edgardashboard.xbrlcloud.com/edgar-dashboard/dashboard.do
> > >
> > > That is just scratching the surface.
> > >
> > > I am not sure you are seeing all the different things that the
> > > model/reference implementation is achieving.  There is a lot going on.
> > >
> > > My model is simple, but it is not simplistic.  You say my model is
> > > not "sufficiently detailed".  What specifically can you show that is
> missing?
> > >
> > > I have run every SEC XBRL financial filing through my model for each
> > > of
> > the
> > > 8000 or so SEC filers.  Works fine.  My model does not detail the
> > > business rules currently, but I have that infoset also.
> > >
> > > Cheers,
> > >
> > > Charlie
> > >
> > > -----Original Message-----
> > > From: Benedikt Kämpgen [mailto:kaempgen@fzi.de]
> > > Sent: Monday, May 07, 2012 7:20 AM
> > > To: 'Charles Hoffman'; 'Benedikt Kämpgen'; 'Frankel, David';
> > > public-gld- comments@w3.org
> > > Cc: 'O'Riain, Sean'; 'Andreas Harth'; 'Herm Fischer'
> > > Subject: RE: Feedback on Ingredients for High Quality Linked Data
> > > section
> > of
> > > Linked Data Cookbook
> > >
> > > Dear Charlie,
> > >
> > > I had found these links before, but thank you very much for putting
> > > them into context for me.
> > >
> > > Regarding your reference implementation of a SEC filing, including
> > business
> > > rules etc. [1]: This is a comprehensive description in HTML and XML
> > > of how one valid filing can be composed into its many parts, and
> > > validated. I am wondering whether there is a description (or even
> > > software) to
> > automatically
> > > retrieve and validate this kind of information from any filing. If
> > > there
> > were, I
> > > could try to understand it and try to do a mapping to a Linked Data
> > > representation.
> > >
> > > Regarding your efforts in aligning XBRL with a common
> > > multidimensional model
> > > [2]: Your ontology [3] is a nice start but I think that filings
> > > cannot be sufficiently detailed described with it to be of use for
> applications.
> > XBRL
> > > Dimensions (Herm Fischer) [4] and the XBRL Abstract Model [5] (David
> > > Frankel) go into much more detail, and I see a lot of
> > > correspondences to
> > the
> > > RDF Data Cube Vocabulary (QB); yet, making the models work together
> > > technically and semantically, would require some thinking.
> > >
> > > Regarding use cases: The standardization of a Linked Data vocabulary
> > typically
> > > is mainly driven by requirements and issues that have been derived
> > > from concrete use cases. Just "aligning XBRL and QB" might be too
> > > fuzzy. Our
> > XBRL
> > > submission [6] might describe a use case, but I guess, the more
> > real-world-
> > > motivated a use case, the better.
> > >
> > > Best,
> > >
> > > Benedikt
> > >
> > >
> > > [1] <http://www.xbrlsite.com/2012/ReferenceImplementation/2012-04-
> > > 15/>
> > > [2]
> > > <http://digitalfinancialreporting.wikispaces.com/Multidimensional+Mo
> > > de
> > > l>
> > > [3]
> > > <http://www.xbrlsite.com/DigitalFinancialReporting/Ontologies/Multid
> > > im
> > > en
> > > sion
> > > alModel.xml>
> > > [4]
> > > <http://www.xbrl.org/Specification/XDT-REC-2006-09-18+Corrected-Erra
> > > ta
> > > -
> > > 2009-
> > > 09-07.htm>
> > > [5]
> > > <http://xbrl.org/Specification/abstractmodel-primary/PWD-2011-10-
> > > 19/abstract
> > > model-primary-PWD-2011-10-19.html>
> > > [6] <http://xbrl.us/research/appdev/pages/275.aspx#>
> > >
> > > --
> > > AIFB, Karlsruhe Institute of Technology (KIT)
> > > Phone: +49 721 608-47946
> > > Email: benedikt.kaempgen@kit.edu
> > > Web: http://www.aifb.kit.edu/web/Hauptseite/en
> > >
> > >
> > >
> > > > -----Original Message-----
> > > > From: Charles Hoffman [mailto:CharlesHoffman@olywa.net]
> > > > Sent: Wednesday, May 02, 2012 3:35 PM
> > > > To: 'Benedikt Kämpgen'; 'Frankel, David';
> > > > public-gld-comments@w3.org
> > > > Cc: 'O'Riain, Sean'; 'Andreas Harth'; 'Herm Fischer'
> > > > Subject: RE: Feedback on Ingredients for High Quality Linked Data
> > > > section
> > > of
> > > > Linked Data Cookbook
> > > >
> > > > Benedikt
> > > >
> > > > In terms of the difficulties in retrieving all linkbase
> > > > information from
> > > the
> > > > filings and taxonomies; I have distilled SEC filings down to a set
> > > > of easy
> > > to
> > > > understand infosets.  The infosets represent the data model
semantics.
> > > > You can get all this information here:
> > > >
> > > > http://www.xbrlsite.com/2012/ReferenceImplementation/2012-04-15/
> > > >
> > > > In particular, this is the fact table infoset:
> > > >
> > > > XML:
> > > > http://www.xbrlsite.com/2012/ReferenceImplementation/2012-04-
> > > 15/abc-
> > > > 20101231
> > > > _FactTable_SEC.xml
> > > > HTML:
> > > > http://www.xbrlsite.com/2012/ReferenceImplementation/2012-04-
> > > > 15_Verification
> > > > /Viewer.html (Go to the individual "Fact Tables" for the
components).
> > > >
> > > > This is the relations infoset:
> > > >
> > > > XML:
> > > > http://www.xbrlsite.com/2012/ReferenceImplementation/2012-04-
> > > 15/abc-
> > > > 20101231
> > > > _Relations_SEC.xml
> > > > HTML:
> > > > http://www.xbrlsite.com/2012/ReferenceImplementation/2012-04-
> > > 15/abc-
> > > > 20101231
> > > > _Relations.html
> > > >
> > > >
> > > > There is a lot more to all this, all is explained here if you are
> > > > interested:
> > > >
> > > > http://xbrl.squarespace.com/digital-financial-reporting/
> > > >
> > > > Basically, you can work with all this "stuff" semantically at the
> > > > report
> > > level.
> > > > No need to even understand the XBRL technical syntax, focus on the
> > > > report semantics.  Dave Frankel is developing something similar in
> > > > a more
> > > official
> > > > format.  But, I know my model works; it is tested against
> > > > thousands and thousands of SEC filings.  If my model works with
> > > > your model I am quite confident that your model, my model, Dave's
> > > > model, XBRL, can be made 100% interoperable.
> > > >
> > > > If you need any additional information of test cases, be sure to
> > > > let me
> > > know.
> > > > I don’t know what you desire in terms of a business use case, but
> > > > I have business use cases "coming out of my ears".
> > > >
> > > > Cheers,
> > > >
> > > > Charlie
> > > >
> > > > -----Original Message-----
> > > > From: Benedikt Kämpgen [mailto:kaempgen@fzi.de]
> > > > Sent: Wednesday, May 02, 2012 2:32 AM
> > > > To: 'Frankel, David'; Charles Hoffman; 'Benedikt Kämpgen';
> > > > public-gld- comments@w3.org
> > > > Cc: 'O'Riain, Sean'; 'Andreas Harth'; Herm Fischer
> > > > Subject: RE: Feedback on Ingredients for High Quality Linked Data
> > > > section
> > > of
> > > > Linked Data Cookbook
> > > >
> > > > Dear Charles, Dear David,
> > > >
> > > > Thanks for your messages and your exciting thoughts about making
> > > > the RDF Data Cube vocabulary (QB), XBRL, and CWM interoperable.
> > > >
> > > > > What I believe would be a good thing is if your data cube,
> > > > > XBRL's data
> > > > cube,
> > > > > and the CWM data cube were 100% semantically interoperable.
> > > > I would agree. But there is probably some alignment work to do:
> > > > For instance, when translating XBRL into QB [1] we had
> > > > difficulties in
> > > retrieving all
> > > > linkbase information from the filings and taxonomies (e.g.,
> > > > concept hierarchies and calculation arcs) and representing them in
> RDF.
> > > > Also, I am aware of the CWM multidimensional model and would be
> > > > interested in how it can be extended for sharing multidimensional
> data.
> > > >
> > > > > other data stores and metadata to XBRL.  I'm cc'ing Herm
> > > > > Fischer, who is a key person in our current effort to raise the
> > > > > level of abstraction at
> > > > which
> > > > > XBRL report design and creation operates, and who is doing some
> > > > > prototyping around the use of linked data with XBRL.
> > > > Since we have been working on using XBRL with Linked Data [1], I
> > > > would be happy to give feedback on your prototypes.
> > > >
> > > > > Formula).  Business rules provides important functionality to
> > > > > the sorts of things XBRL does with financial reporting (making
> > > > > sure the information is
> > > > > correct) and I believe that this same functionality is necessary
> > > > > for
> > > > quality
> > > > > business reporting of any kind; financial, nonfinancial,
> > > > > government, industry, anything.
> > > > Regarding business rules, we are considering refining QB to more
> > > > formally define relationships between data cubes (e.g.,
> > > > aggregations), which may help to represent more complex formulae
> > > > or business rules using
> > > QB data.
> > > >
> > > > > The question is how to put all these things together?
> > > > One way to start this effort would be to define a use case around
> > > > XBRL, QB and possibly CWM, which could help to derive requirements
> > > > for them to become interoperable. Some proposed use cases for QB
> > > > are available
> > > at [2].
> > > > Their technical detail varies, but in general they should give an
> > > > overview
> > > of
> > > > what QB (e.g., together with XBRL, CWM) may allow to do.
> > > >
> > > > Best,
> > > >
> > > > Benedikt
> > > >
> > > > [1] <http://xbrl.us/research/appdev/pages/275.aspx#>
> > > > [2]
> > > <http://www.w3.org/2011/gld/wiki/Data_Cube_Vocabulary/Use_Cases>
> > > >
> > > >
> > > >
> > > > --
> > > > AIFB, Karlsruhe Institute of Technology (KIT)
> > > > Phone: +49 721 608-47946
> > > > Email: benedikt.kaempgen@kit.edu
> > > > Web: http://www.aifb.kit.edu/web/Hauptseite/en
> > > >
> > > >
> > > >
> > > >
> > > >
> > > > > -----Original Message-----
> > > > > From: Frankel, David [mailto:david.frankel@sap.com]
> > > > > Sent: Thursday, April 26, 2012 7:13 PM
> > > > > To: Charles Hoffman; 'Benedikt Kämpgen';
> > > > > public-gld-comments@w3.org
> > > > > Cc: 'O'Riain, Sean'; 'Andreas Harth'; Herm Fischer
> > > > > Subject: RE: Feedback on Ingredients for High Quality Linked
> > > > > Data section
> > > > of
> > > > > Linked Data Cookbook
> > > > >
> > > > > Thanks, Charlie.  I consider linked data to be important for
> > > > > semantic federation of the various XBRL taxonomies and report
> > > > > data, and for linking other data stores and metadata to XBRL.
> > > > > I'm cc'ing Herm Fischer, who is a key person in our current
> > > > > effort to raise the level of abstraction at
> > > > which
> > > > > XBRL report design and creation operates, and who is doing some
> > > > > prototyping around the use of linked data with XBRL.
> > > > >
> > > > > Regards,
> > > > >
> > > > > --David
> > > > >
> > > > > David S. Frankel
> > > > > Standards and Open Source Strategy Technology and Innovation
> > > > > Platform Group SAP Labs LLC; Palo Alto, California USA Phone &
> > > > > Cell +1 530 591-0212
> > > > > Email: david.frankel@sap.com
> > > > >
> > > > >
> > > > > -----Original Message-----
> > > > > From: Charles Hoffman [mailto:CharlesHoffman@olywa.net]
> > > > > Sent: Thursday, April 26, 2012 8:51 AM
> > > > > To: 'Benedikt Kämpgen'; public-gld-comments@w3.org
> > > > > Cc: 'O'Riain, Sean'; 'Andreas Harth'; Frankel, David
> > > > > Subject: RE: Feedback on Ingredients for High Quality Linked
> > > > > Data section
> > > > of
> > > > > Linked Data Cookbook
> > > > >
> > > > > Benedikt;
> > > > >
> > > > > (I added David Frankel to this list, you will see why in a
> > > > > moment as you read. David, you will likewise understand, please
> > > > > read this
> > > > > thread.)
> > > > >
> > > > > Your RDF Data Cube Vocabulary is right on target in my view.
> > > > > There is a group within the XBRL community which is modeling
> > > > > something similar for XBRL. David Frankel is leading that effort.
> > > > > David can explain this better, but in short as I understand it;
> > > > > XBRL is trying to leverage the work of
> > > > the
> > > > > on CWM, Common Warehouse Metamodel:
> > > > > http://www.omg.org/spec/CWM/
> > > > >
> > > > > What I believe would be a good thing is if your data cube,
> > > > > XBRL's data
> > > > cube,
> > > > > and the CWM data cube were 100% semantically interoperable.
> > > > >
> > > > > The second thing is that XBRL has a business rules language
> > > > > (XBRL
> > > > > Formula) and business rules engines (part of an XBRL processor
> > > > > which implements XBRL Formula).  Business rules provides
> > > > > important functionality to the sorts of things XBRL does with
> > > > > financial reporting (making sure the information is
> > > > > correct) and I believe that this same functionality is necessary
> > > > > for
> > > > quality
> > > > > business reporting of any kind; financial, nonfinancial,
> > > > > government, industry, anything.
> > > > >
> > > > > The question is how to put all these things together?  We have
> > > > > the Government Linked Data Working Group, Data Transparency
> > > > > Coalition
> > > > > (http://datacoalition.org/) pushing on the DATA Act
> > > > > (http://keepthewebopen.com/data).  We have the ADMS folks in
> > > > > Europe pushing more on semantics
> > > > > (https://joinup.ec.europa.eu/asset/adms/home) and I am sure I
> > > > > left a few out.
> > > > >
> > > > > How do we make all these things "play well" together and serve
> > > > > government and business well, globally?
> > > > >
> > > > > So basically, that is my view.  XBRL and your RDF Data Cube
> > > > > Vocabulary should be interoperable.
> > > > >
> > > > > I am sure you and David will have a lot to talk about!
> > > > >
> > > > > Cheers,
> > > > >
> > > > > Charlie
> > > > >
> > > > > -----Original Message-----
> > > > > From: Benedikt Kämpgen [mailto:kaempgen@fzi.de]
> > > > > Sent: Thursday, April 26, 2012 2:41 AM
> > > > > To: 'Charles Hoffman'; public-gld-comments@w3.org
> > > > > Cc: O'Riain, Sean; 'Andreas Harth'
> > > > > Subject: RE: Feedback on Ingredients for High Quality Linked
> > > > > Data section
> > > > of
> > > > > Linked Data Cookbook
> > > > >
> > > > > Dear Charles Hoffman,
> > > > >
> > > > > Thanks for your recommendation to look more closely into XBRL,
> > > > > for instance, w.r.t. the application of business rules on Linked
> Data.
> > > > >
> > > > > What may be of interest in this regard:
> > > > >
> > > > > 1) In GLD, we are considering to add XBRL use cases to the
> > > > > development of the "RDF Data Cube Vocabulary" (QB) for
> > > > > publishing statistics (such as financial disclosures). See [1]
> > > > > for the current
> > > vocabulary QB.
> > > > > Examples of possible use cases:
> > > > >
> > > > > 1.1)  Colleagues and I have used QB at [2] (an XBRL challenge
> > > > > submission)
> > > > to
> > > > > publish XBRL filings from SEC as Linked Data and to consume
> > > > > those filings using Online Analytical Processing.
> > > > >
> > > > > 1.2) At [3] of our current QB use case document (not official,
> > > > > reviewed or published), a possible use case (UC 10) is described
> > > > > that transforms financial statistics as Linked Data reusing QB
> > > > > into
> > > XBRL.
> > > > >
> > > > > 2) There will be a paper on using rules on financial Linked Data
> > > > > (XBRL as Linked Data) in the FEOSW workshop at ESWC 2012 [4]
> > > > >
> > > > > I would be happy to hear your opinion about using XBRL in use
> > > > > cases of the RDF Data Cube Vocabulary.
> > > > >
> > > > > Best,
> > > > >
> > > > > Benedikt
> > > > >
> > > > > [1] <http://www.w3.org/TR/2012/WD-vocab-data-cube-20120405/>
> > > > > [2] <http://xbrl.us/research/appdev/pages/275.aspx#>
> > > > > [3]
> > > > > <http://dvcs.w3.org/hg/gld/raw-file/default/data-cube-
> > > > > ucr/index.html#transfo
> > > > > rming-published-statistics-into-xbrl--uc-10>
> > > > > [4] <http://nadir.uc3m.es/feosw2012/#ui-tabs-9>
> > > > > Title: Using SPIN to Formalise Accounting Regulations on the
> > > > > Semantic Web
> > > > >
> > > > > Authors: Dennis Spohr, Philipp Cimiano, John McCrae and Seán
> > > > > O'Riain
> > > > >
> > > > > Abstract:
> > > > > The eXtensible Business Reporting Language (XBRL) has
> > > > > standardised financial reporting and provide a
> > > > > machine-interpretable format that makes financial and business
> > > > > reports easier to access and
> > consume.
> > > > > Leveraging XBRL with Open Linked Data for purposes such as
> > > > > multi- dimensional regulatory querying and investigation
> > > > > requires XBRL formulisation as RDF.
> > > > > This paper investigates the use of of-the-shelf Semantic Web
> > > > > technologies
> > > > to
> > > > > formulise accounting regulations specified in XBRL
> > > > > jurisdictional taxonomies. Specifically the use of the SPARQL
> > > > > Inferencing Notation
> > > > > (SPIN) with RDF to represent these accounting regulations as
> > > > > rule constraints,
> > > > not
> > > > > cater for in the RDF abstract model is investigated. We move
> > > > > beyond previous RDF to XBRL transformations and investigate how
> > > > > SPIN enhanced formalisation enables inferencing of financial
> > > > > statement facts associated with financial reporting concepts and
> > > > > sophisticated consistency checks, which evaluate
> > > > the
> > > > > correctness of reported financial data with respect to the
> > > > > calculation requirements imposed by accounting regulation. The
> > > > > approach illustrated through two use cases demonstrates the use
> > > > > of SPIN to meet central requirements for financial data and
> > > > > regulatory
> > > modelling.
> > > > >
> > > > >
> > > > > --
> > > > > AIFB, Karlsruhe Institute of Technology (KIT)
> > > > > Phone: +49 721 608-47946
> > > > > Email: benedikt.kaempgen@kit.edu
> > > > > Web: http://www.aifb.kit.edu/web/Hauptseite/en
> > > > >
> > > > >
> > > > > > -----Original Message-----
> > > > > > From: Charles Hoffman [mailto:CharlesHoffman@olywa.net]
> > > > > > Sent: Monday, April 16, 2012 4:28 PM
> > > > > > To: public-gld-comments@w3.org
> > > > > > Subject: Feedback on Ingredients for High Quality Linked Data
> > > > > > section of Linked Data Cookbook
> > > > > >
> > > > > > This is a great working group, nicely organized.  In
> > > > > > particular the Linked
> > > > > Data
> > > > > > Cookbook is quite useful.
> > > > > >
> > > > > >
> > > > > >
> > > > > > I do have some feedback for the High Quality Linked Data
> > > > > > section of that Linked Data Cookbook.
> > > > > >
> > > > > >
> > > > > >
> > > > > > It is my view that one thing missing from the list of items
> > > > > > necessary for
> > > > > high
> > > > > > quality linked data is business rules.  In particular
> > > > > > computations or
> > > > > relations
> > > > > > between information items.
> > > > > >
> > > > > >
> > > > > >
> > > > > > This is a very good summary/example of what I am taking about
> > > > > > and way I have this position:
> > > > > >
> > > > > >
> > > > > >
> > > > > > http://xbrl.squarespace.com/journal/2010/5/27/census-bureau-co
> > > > > > nf
> > > > > > ir
> > > > > > ms
> > > > > > -
> > > > > > error-revises-data-set.html
> > > > > >
> > > > > >
> > > > > >
> > > > > > The cliff notes are that the US Census Bureau published data,
> > > > > > the format
> > > > > was
> > > > > > CSV.  If the data were in RDF, the same issue would exist.
> > > > > > The data had
> > > > > an
> > > > > > error in it.  It was not until I created business rules to be
> > > > > > sure that my
> > > > > use of
> > > > > > the data was correct that I discovered an error in the US
> > > > > > Census
> > data.
> > > > > > I
> > > > > found
> > > > > > this error because I wanted to be sure the XBRL information I
> > > > > > was creating was correct. As such, I created business rules,
> > > > > > using XBRL, to verify that
> > > > > my
> > > > > > data set was correct.  And that is how I found the error.
> > > > > >
> > > > > >
> > > > > >
> > > > > > Said another way, if the data set had business rules provided
> > > > > > WITH the
> > > > > data
> > > > > > set, then (a) those providing the data would have become aware
> > > > > > of the
> > > > > error
> > > > > > and (b) those using the data could both better understand the
> > > > > relationships
> > > > > > because they are articulated and they can validate the
> > > > > > information prior
> > > > > to
> > > > > > use to confirm that there is no error.
> > > > > >
> > > > > >
> > > > > >
> > > > > > Thank you for considering this feedback.
> > > > > >
> > > > > >
> > > > > >
> > > > > > Cheers,
> > > > > >
> > > > > >
> > > > > >
> > > > > > Charles Hoffman, CPA
> > > > >
> > > > >
> > > >
> > > >
> > >
> > >
> >
> >
> 
> 

Received on Tuesday, 22 May 2012 20:42:08 UTC