Re: About presentation at SWIG meetin in technical plenary 2006

Dear Yasuhara-san,

* yasuhara@instsec.org <yasuhara@instsec.org> [2006-02-10 22:28+0900]

> This mail is sent by the suggestion of Mr.R. Swick(W3C). I am a researcher 
> engaged in ISeC.
 
Many thanks for bringing this work to the Interest Group's attention. It
seems to connect with W3C Semantic Web efforts in a number of ways.

> Institute of Semantic Computing ISeC is non-profit organization settled by 
> Japanese academic research members. Our first public action was The Semantic 
> Computing Initiative SeC2006 workshop held at WWW2005 in Makuhari Japan. 
> http://www.instsec.org/2005ws/
> Paper title:" CDL (Concept Description Language):A Common Language for 
> Semantic Computing"
> http://www.instsec.org/2005ws/papers/yokoi.pdf ;

I've not studied this carefully enough yet, but my first impression is that
CDL goes a lot further in the direction of natural language
representation than any existing W3C Semantic Web technologies. There is
some (relatively early) work at W3C on the representation of Wordnet
within RDF, but this only addresses the English version of Wordnet. See 
http://lists.w3.org/Archives/Public/public-swbp-wg/2006Feb/0009.html for 
pre-publication work on Wordnet-in-RDF. CDL seems to be a lot more general
and ambitious in scope. There is also I think some connection to 
W3C's SKOS work, which (like CDL) models conceptual structures, and
which may be extended eventually to address more lexical/linguistic
aspects too. See http://www.w3.org/2004/02/skos/ for more on SKOS.

In section 4 of http://www.instsec.org/2005ws/papers/yokoi.pdf I read 

[[
The design principle of RDF+OWL is to describe property of resource. On 
the other hand, that of CDL is to describe conceptual structure of
objective content and world. Content and world are also regarded as
resource. Therefore the function of RDF+OWL is to describe property of 
content and world, that is, to make metadata of content and world 
machine-understandable. On the other hand, CDL has a purpose of making
concept structure of content and world machine-understandable. However 
structure description and property description are utterly different
but complementary as well. Therefore CDL and RDF-OWL are utterly
different and complementary. Accordingly, in developing the
specification of CDL, similar parts in the specification of RDF-OWL 
is utilized as much as possible and the implementation method on RDF+OWL 
is also proposed.
]]


This is an issue that we've encountered several times in the Interest
Group --- how representational systems that are similar to RDF,
but address different priorities (and sometimes with different expressivity)
can interoperate with RDF. TopicMaps are another example, as are rule 
languages.

A related question... I'm curious about how CDL relates to UNL
(http://www.undl.org/). Your paper mentions a CDL version of UNL. 
Are the two systems being developed in  parallel? Do they address similar 
issues and use cases? I learnt a little about UNL at the recent UN World 
Summit on the Information Society in Tunis, and would be interested to 
hear more about the relationship between the two.

> CDL R&D is supported by three years Japanese government fund from Ministry of 
> Internal Affairs and Communications (MIC) since FY2005. The mission of the CDL 
> R&D is to promote international standardization of CDL. Our plan is to submit 
> CDL aiming at recommendation from W3C because ISeC considers CDL to be applied 
> to Web content description. 

I'm very happy to see discussion of CDL and its relationship to RDF. I expect
we would need to do quite a lot more work to understand the relative
scope of CDL and RDF, and the potential for expressing CDL in RDF. My
own personal view is that there is definitely a role for standardisation 
work on top of RDF that makes linguistic and conceptual structures 
accessible in the Semantic Web, so perhaps CDL.nl in particular would be 
a good candidate for standardisation. There is also work in other areas
of W3C, eg. the Voice Browser effort, http://www.w3.org/Voice/ ...that
might need to be taken into account. For example, there is work on 
representing pronunciation information: 
http://www.w3.org/TR/2006/WD-pronunciation-lexicon-20060131/ as well as 
a discussion of new "incubator" work on multimedia descriptions, which
relates to your MPEG use cases. If you're interested in that, se the 
charter drafting discussion at 
http://lists.w3.org/Archives/Public/public-swbp-wg/2006Feb/0068.html


>			During the W3C technical plenary week 2006 March 
> 2nd, ISeC expects your Semantic Web Interest Group will approve to make 
> presentation in the following:
> 
> Title: ?$B!HCDL as a basis for World Common Web Language?$B!I
> By Hiroshi Uchida*, Meiying Zhu, Hiroshi Yasuhara, Toshio Yokoi (ISeC)
> Outline: W3C Semantic Web Activity recommended Web Resource Description 
> Framework and Web Ontology Language. Semantic description of Web resources 
> will proceed from metadata towards content itself. The vision of Concept 
> Description Language CDL.* is to describe multi media content uniformly, whose 
> data model is semantic network with hyper graph structure. This talk will 
> focus on CDL.nl. CDL.nl is specified as a language for natural language media. 
> RDF has triple data model representing subject, predicate and object. CDL.nl 
> represents exactly syntactic structure of natural language. The presentation 
> shows syntax and semantics of CDL.nl. CDL.nl equipped with lexical concept or 
> ontology is a computer language equivalent for natural language 
> representation. Conversion of CDL.nl and lexical concept to RDF and OWL will 
> derive a ?$B!HWorld Common Web Language?$B!I on the semantic web schema.
> I am looking forward to receiving your response.

Unfortunately I won't be at the meeting myself. Danny Ayers (cc:'d) is 
chairing the meeting. I've just talked with Danny. The agenda is now 
quite full ( draft at http://esw.w3.org/topic/SwigAtTp2006 ), but he
is confident that there is space for a 15 minute CDL / CDL.nl presentation 
followed by some questions and discussion time. This should be enough time 
to introduce the main issues and identify common interests that can be 
followed up in more detail afterwards. Danny is handling the specifics of 
the meeting time allocations, so please keep him Cc:'d in any correspondence.

Thanks again for raising this topic, and for any further thoughts you 
might have on the relationship of CDL to the other W3C work I mentioned.

cheers,

Dan

ps. are there any publically available CDL datasets that could be 
used for RDF experimentation?

> Best regards,
> 
> Hiroshi Yasuhara
> ISeC

----- End forwarded message -----

Received on Friday, 10 February 2006 16:46:38 UTC