W3C home > Mailing lists > Public > semantic-web@w3.org > April 2008

Re: [Semantic_Web] Re: SemWeb in Real Life - ie not just theory

From: Danny Ayers <danny.ayers@gmail.com>
Date: Wed, 9 Apr 2008 11:05:32 +0200
Message-ID: <1f2ed5cd0804090205s6bba5f42p9b1d0e022bcfdc30@mail.gmail.com>
To: semantic_web@googlegroups.com, SWIG <semantic-web@w3.org>
[cc'ing semantic-web@w3.org, original thread at
http://groups.google.com/group/semantic_web/browse_thread/thread/299e6336e05f5188]

On 05/04/2008, Logician <sales@logicians.com> wrote:
>
>
> The issue of language, or semantics, in my business has a different
> meaning to just understanding words. The difference is that products
> are described by design criteria, quality, etc. It is not a matter of
> understanding English words and translating into a machine language
> (eg SQL) but of actually understanding the terms and the data related
> to them, almost like a new language.


That actually sounds the kind of job for which RDF and  OWL (the Web
Ontology Language) are well suited, though offhand I can't think of any
existing vocabularies in that space.

The key difference is that in sales qualitative issues matter. There
> are few facts about a product. If I say High Quality it is not a fact
> but a view. If several people have the view, it is an accepted view. A
> parallel would be to invent a language to describe something because
> words do not exist. Inventing the language would be hard but once
> invented it could be used to describe all similar objects.


Right, the nearest thing I know of to that is the Review vocabulary:

http://purl.org/stuff/rev#

which is used under the hood at http://revyu.com



I agree once that is done, you could use your kind of thinking to them
> find different ways of saying High Quality, eg good products,
> recommended, etc
>
> To clarify: Microsoft invested in supported XML with classes in C#
> making it easy to use XML. Now we have RDF and recommendations to
> extend XML, I assume Microsoft will invest in more C# classes so we
> can easily process RDF.


Microsoft haven't exactly been quick off the mark for RDF support (though
their ADO.Net Entity Framework stuff is very RDF-like, but unfortunately
doesn't use URIs as identifiers). Fortunately there are loads of open source
toolkits available, see:
http://www4.wiwiss.fu-berlin.de/bizer/toolkits/

I think once key difference is that semantics related to sales is a
> business, and not actually overhead (eg accounts is overhead). The key
> to distribution is product knowledge. If such knowledge can be changed
> then the entire retail systems of the world can change allowing people
> to find what they want almost instantly. This has an impact of
> billions as billions are used in advertising, brochures, shop
> displays, sales calls, branding, etc.


Absolutely!

Not long ago I started looking into product description for a considerably
simpler scenario than you describe (for music equipment, the main
complication was representing part-whole relationships for guitar parts
etc). I was surprised to find how little work had apparently been done in
this area, given the commercial potential.

For what it's worth I got as far as thinking something very like the core
terms of FRBR would make a good starting point, borrowing the part-whole
stuff from somewhere like Cyc or maybe one of the proposed Upper Ontologies,
the whole lot being liberally sprinkled with SKOW.

Cheers,
Danny.

-- 
http://dannyayers.com
~
http://blogs.talis.com/nodalities/this_weeks_semantic_web/
Received on Wednesday, 9 April 2008 09:12:27 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 07:42:03 UTC