W3C home > Mailing lists > Public > public-sws-ig@w3.org > October 2005

Re: LAM technologies for SW -> FCS-Ontologies [PUBLIC-SWS-IG]

From: hj nern [I] <nern@infrawebs.info>
Date: Sat, 08 Oct 2005 13:10:25 +0200
Message-ID: <4347A921.4050307@infrawebs.info>
To: Shashi Kant <shashi.mit@gmail.com>
CC: Linda Ridgeway <linda@alierra.com>, public-sws-ig@w3.org

Hi Linda, Hi Shashi

Shashi Kant wrote:
> Hi Linda
> 
> There is plenty of interest in this field and it goes by several names: 
> Bayesian reasoning, uncertainty reasoning, statistical 
> reasoning/inferencing and what have you....
> OWL/RDF(s) inferencing are a (deterministic) subset of uncertainty 
> reasoning and combing these two is also the thrust of my research at MIT.

Nice to hear that you re busy with this topics.
It s obvious that the current US QWL/RDF(s) as well as the EU WSMO 
stream deal "only" with hard semantics in the description and 
manipulation of crisp data.
They do not have the capability to process soft semantics.
As long as it will be used bivalent-based logical methods there will be 
no progress in handling of ill-structured, uncertain and imprecise 
information, frequently encountered in real world knowledge.
The forementioned initiatives do not consider sufficiently the 
implicitly given non-precision and non-accuracy of real world scenarios 
and applications.
Further it has to be observed that for the evaluation of developed 
structures and environments only very specific test beds and use cases 
are chosen, which provide salutary and convenient limitations in the 
sense of a deformation of “real world” conditions to “special world” 
conditions.

In this context I d like to give a pointer to the European stream around 
Elie Sanchez (FCS-O -> Fuzzy Concept Set Ontologies, "Fuzzy Logic and 
the Semantic Web") from Marseille and Umberto Straccia from Pisa ("Fuzzy 
Description Logic for the Semantic Web"). They show up convincing 
alternative methods and techniques out of the domain of soft-computing.

Further Remark: Bayesian Reasoning is not "always" stable and convergent 
- or better it converges very "slowly" and sometimes not (depends on the 
underlying BN). But Fuzzy inferencing over FCS-O provides well known and 
approved techniques and methods - and guarantees a stable reasoning.

BR & HTH
  hjn

> 
> If you are interested in some further reading along this topic, a superb 
> book is "On Intelligence" by Jeff Hawkins. Jeff was the co-founder of 
> Palm and has now founded a company called Numenta to work in this field. 
> It is meant for non-technical audience and is a fantastic read for 
> anyone remotely interested in this topic.

I wish a happy and eager studying.

> 
> HTH,
> Shashi
> 



-- 

____________________________________________________
HJ NERN INFRAWEBS (IST FP6-511723)

mail:    sci.coord@infrawebs.org
net:     http://www.infrawebs.org
net:     http://www.aspasia-systems.de

____________________________________________________

This e-mail may contain confidential and/or privileged information. If 
you are not the intended recipient (or have received this e-mail in 
error) please notify the sender immediately and destroy this e-mail. Any 
unauthorized copying, disclosure or distribution of the material in this 
e-mail is strictly forbidden.
Received on Saturday, 8 October 2005 12:20:12 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Sunday, 16 March 2008 00:11:02 GMT