+1
(01)
Sjir Nijssen
(02)
Chief Technical Officer
PNA Group
(03)
Tel: +31 (0)88-777 0 444
Mob: +31 (0)6-21 510 844
Fax: +31 (0)88-777 0 499
E-mail: sjir.nijssen@xxxxxxxxxxxxx
-------------------------------------------------------
http://www.pna-group.com
(04)
________________________________________
Van: ontolog-forum-bounces@xxxxxxxxxxxxxxxx
[ontolog-forum-bounces@xxxxxxxxxxxxxxxx] namens Paul Tyson
[phtyson@xxxxxxxxxxxxx]
Verzonden: zondag 23 juni 2013 7:43
To: [ontolog-forum]
Onderwerp: Re: [ontolog-forum] RDF and XML
(05)
On Sat, 2013-06-22 at 11:30 -0500, Pat Hayes wrote:
> Ahem.
>
> As someone who is heavily involved in both the CL and RDF 1.1
> communities, I have to say that I find this idea - of basing SWeb
> formalisms on Common Logic - completely unworkable. Not for technical
> reasons, but for political/social ones. There is absolutely no way
> that anything as complicated as CL is going to be adopted by Web
> developers and software engineers in the forseeable future. The
> touchstone question to ask about any proposed data model or format is
> this, I have been repeatedly told: Can a software developer use it to
> build a prototype application that does something interesting or
> useful, using easily available tools and systems, in a couple of days,
> without having to read anything more than a few paragraphs of
> explanation or learn more than one new idea?
(06)
Well that's a pretty small box to expect impressive results from. But I
have a case study that would fit in a slightly larger box.
(07)
"Easily available tools and systems":
(08)
xml
html
rdf(s)
RDB2RDF
xslt
sparql
Common Logic (the broken xcl1.0)
emacs (+ nxml-mode)
eclipse
saxon XSLT engine
java
jdbc
javascript
apache httpd
apache tomcat
apache jena
VirtualBox
-------------
Total cost: $0.00
(09)
Time to implement: more than "a couple of days". There are a lot of
moving parts, but no piece took more than a few weeks to get a prototype
running--and some actually were done in days (or hours). We took it
quickly to production scale, so we were doing POC, development, and beta
production support concurrently.
(010)
"Interesting or useful": We're maintaining an RDF repository with
several hundred million triples representing "lifeblood" information to
the business. The repository is stoked with triples mapped from RDBMS
using R2RML. We identify data integrity problems using SPARQL queries
generated from XCL. These are data problems that are difficult or
impossible to detect in the native systems, but will impede business
processes. We provide some linked data browsing capabilities over the
repository, including a showpiece application that demonstrates
unparalleled ability to view data from disparate systems.
(011)
I wanted to put as much business logic as possible in a standard rule
language. The choices appeared to be: 1) RIF; 2) RuleML; and 3) CL. RIF
probably would have worked, but is verbose and unwieldy with its
dialects. Even worse for RuleML. CL seemed simple and flexible by
comparison, and it was easy to get a Relax NG grammar for xcl
straightaway.
(012)
Admittedly, CL is a small piece of the system, but it has proven its
worth. The XCL source is used to generate: three different types of
SPARQL queries (ASK, CONSTRUCT, SELECT); alternative queries returning
different result sets (full or summary); and query templates that can be
filled in with parameters. The XCL is also formatted to HTML so the
rules can be read by business users--removing some of the mystique (and
confusion) that usually surrounds SQL wizardry.
(013)
Future enhancements will enlarge the role of CL to touch more parts of
the system. SQL could be generated from it, to support less-evolved
applications. It could inform client applications. It could be used to
generate or check R2RML mappings. I'm looking forward to the new and
improved features of xcl2.
(014)
So I don't buy the "it's too hard" argument.
(015)
> If not, forget it. No
> matter how wonderful it might be, it will not get taken up and adopted
> by enough developers to make it a viable standard.
(016)
Good. That gives me a competitive advantage.
(017)
>
>
> Let me take as an illustrative example the 2004 RDF Semantics
> specification document. RDF is an almost trivially simple logic
> compared to CL, and its model theory is if anything even simpler by
> comparison (because the CL model theory is actually fairly intricate,
> for technical/historical reasons that need not concern us here.) And
> we wrote that 2004 document deliberately in a kind of 'tutorial' mode
> because we knew that many readers would not be familiar with
> model-theoretic semantics, so it is thick with internal links to a
> thoroughly comprehensive glossary, etc.. Some "logical" readers
> thought it was a really nicely written document. In the Semantic
> Web/Linked Data world of Web developers, it has been the single most
> disastrous cause for the failure of RDF to be adopted. RDF is now
> almost universally perceived as complicated and arcane and generally
> weird, so much so that even to use the three letters R-D-F in the
> introduction to JSON-LD (which is an RDF dialect written in JSON)
> has been resisted as likely to harm adoption. Just using the word
> "semantics" is widely seen in many communities as likely to destroy
> any chances of wide adoption of any technology.
>
> It is no good writing tutorials and glossaries for people who are not
> motivated to learn something new, and developers are not at school.
> They just want to get software working. They don't give a damn about
> expressiveness of logics or about validity of entailments, and they
> *certainly* don't care about anything that smacks of philosophy.
>
(018)
I wouldn't put all the blame on developers, or paint them all with that
critical brush. They do what they are paid to do, and if they don't like
what they're paid to do they'll go work somewhere else. So put half or
more of the blame on the people who employ and direct the developers,
and the strangely regressive management culture that seems to be common
in North American corporate IT departments.
(019)
Regards,
--Paul
(020)
> So, you have to sneak the new ideas into people's heads without them
> realizing it. Telling them to translate everything into ISO-CL is not
> a good way to do that. Giving them a way to use JSON that makes in,
> invisibly, into a fairly compact notation for RDF is a good sneaky
> start. Tweaking the RDF data model so that it becomes a variant of
> Peirce's existential graph notation (see
> http://www.slideshare.net/PatHayes/blogic-iswc-2009-invited-talk)
> would be a good sneaky next step. But all this will have to be done
> with care, and without telling people that the first thing they should
> do is take a graduate course in model theory or quickly brush up on
> ISO Common Logic. Because they simply won't do that, and if they hear
> you saying that, they will stop listening.
>
> On Jun 22, 2013, at 12:48 AM, John F Sowa wrote:
>
> > Pat C, Michel, and Chris,
> >
> > Pat Hayes and Guha defined the LBase for RDF and OWL as a subset
> > of the CL model theory. That means that all the Semantic Web
> > logics can be represented as various subsets of Common Logic.
> >
> > PC
> >> I would like to find a program that actually *translates* OWL into CL.
> >> Does anyone know of such a program?
> >
> > I don't know of any program that does the translation, but Pat Hayes
> > spelled out the details. The rest, as they say, is ASMOP --
> > A Simple Matter Of Programming.
>
> See, John, this is why developers don't listen. They have to actually DO the
>actual programming. If you think they should use ISO-CL, you write the
>freeware code to make their lives easier.
>
> >
> > http://www.ihmc.us/users/phayes/CL/SW2SCL.html
> > Translating Semantic Web languages into Common Logic
> >
> > MD
> >> Is there no path by which CL can be brought into the fold
> >> of Semantic Web technologies?
> >
> > Recommendation: All the SW languages can be treated as dialects
> > of Common Logic.
>
> That is exactly what L-base was all about. It is already out there as a
>published W3C Note, has been there now for nine years. The take-up is exactly
>zero. This ideas does NOT WORK.
>
> > That means that CL is the universal medium that
> > can relate all of them to one another and to many other versions
> > of logic. In particular, SHOE and F-logic, which I mentioned in
> > my previous note, could be specified as CL dialects. CL could
> > be used to relate them to OWL, if needed for some application.
> >
> > MD
> >> Is there a document that can describe the advantages, disadvantages of
>rdf/owl/cl?
> >
> > There are many people who have discussed these issues, and it would be
> > possible to use Pat H's web page as a basis for specifying the details.
> > The basic advantage would be a common model-theoretic semantics that
> > would play the role of SWeLL in Tim B-L's original proposal.
> >
> > MD
> >> Who would champion it through the W3C process?
> >
> > We'd have to find somebody connected with the W3C who understands
> > the issues and would be willing to talk to a lot of people.
>
> Been there, done that. Who do you talk to? TimBL already understands this
>very well, his N3 system has has more-than-FO expressivity for years. Try
>selling this idea to, say, a half-dozen Web developer companies. Then to a few
>hundred developers. Then to a few thousand Web developers. All you will
>generate is email debate/noise.
>
> And to be less cynical for a moment, what actual advantage would it be to be
>using ISO-CL, even if the tools were available? The amount of data out there
>that needs anything beyond OWL expressivity is vanishingly small. Very few
>applications actually perform anything other than the most trivial inferences.
>I have not seen any mash-up application that needed to use, for example,
>deMorgan's law. So why would anyone use this hammer to crack their nuts?
>
> >
> > C Mungal
> >> Advantages/disadvantages: CL is more expressive than RDF or OWL.
> >> RDF and OWL currently have more libraries, tools, support, mindshare,
> >> persistent stores, etc than CL. What's more important for your
>application?
> >
> > You can have both. When you adopt the more expressive CL as the
> > foundation, all the tools for the subset languages become CL tools.
>
> Just calling them "CL tools" doesn't make them interoperate. And they don't
>interoperate for all kind of messy detail reasons as well as the basic
>semantic incompatibilities. You have to get both of them fixed, the principled
>things and the messy details. We spend more time on how to format and type
>character-string literals than almost any other topic in RDF 1.1
>
> Pat
>
> >
> > The Cyc Project, for example, has a very expressive CycL logic, but
> > it uses a large variety of different reasoning methods for different
> > kinds of problems. The users view CycL as a single, unified language.
> > But for any particular problem, the system automatically selects one
> > reasoning method or another as the one that is most suitable for
> > that problem.
> >
> > I discuss some of these issues in the following article:
> >
> > http://www.jfsowa.com/pubs/fflogic.pdf
> > Fads and Fallacies About Logic
> >
> > John
> >
> > _________________________________________________________________
> > Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
> > Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
> > Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
> > Shared Files: http://ontolog.cim3.net/file/
> > Community Wiki: http://ontolog.cim3.net/wiki/
> > To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
> >
> >
>
> ------------------------------------------------------------
> IHMC (850)434 8903 or (650)494 3973
> 40 South Alcaniz St. (850)202 4416 office
> Pensacola (850)202 4440 fax
> FL 32502 (850)291 0667 mobile
> phayesAT-SIGNihmc.us http://www.ihmc.us/users/phayes
>
>
>
>
>
>
> _________________________________________________________________
> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
> Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
> Shared Files: http://ontolog.cim3.net/file/
> Community Wiki: http://ontolog.cim3.net/wiki/
> To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
>
(021)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J (022)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J (023)
|