- From: Trevor Martin <Trevor.Martin@bristol.ac.uk>
- Date: Thu, 19 Jul 2007 09:22:12 +0100
- To: Giorgos Stoilos <gstoil@image.ece.ntua.gr>
- Cc: "'Umberto Straccia'" <umberto.straccia@isti.cnr.it>, <public-xg-urw3@w3.org>
This is precisely the choice faced by implementers of logic programming + uncertainty languages .... you can extend the language and the inference mechanism or express and process the uncertainty within the standard language. tall(John) : 0.7 vs tall(John, 0.7) (... in both cases, without saying what 0.7 represents) The former approach gives you more control, reduces to "standard" notation when the uncertainty is omitted and (I think) makes the semantics clearer; the latter involves no change to existing notation (hence is easier to sell ) but gets messy when only some of the representation requires the uncertainty and obscures the meaning of the annotation. Trevor (PS apologies for the fact that I have been quiet so far, fuzz-ieee next week has taken a lot more of my time than expected) On 19 Jul 2007, at 08:43, Giorgos Stoilos wrote: > > > >> -----Original Message----- >> From: public-xg-urw3-request@w3.org [mailto:public-xg-urw3- >> request@w3.org] >> On Behalf Of Umberto Straccia >> Sent: Wednesday, July 18, 2007 8:55 PM >> To: public-xg-urw3@w3.org >> Subject: Re: [URW3 public] OWL extensions [was Re: [URW3] ... three >> questions based on the last telecon] >> >> >> I am more in favor of having no OWL/RDFS language extension at all, >> but encoding the data, according to a core uncertainty ontology, >> which we may develop as outcome (see my previous email.... or just >> below). To my opinion, this is by far much more extensible/flexible >> than any OWL/RDFS language extension and certainly faster and easier >> to be accepted by any non-uncertainty community. >> > > My point was *not* that your proposal is not feasible. To the > contrary it is > very nice. My point was that at the end of the day it will turn our > to be > almost (and maybe exactly) the same thing to agree upon using an > OWL concept > like LeftShoulderFunction for left shoulder functions, which crisp > tools > will have to learn to toss away while fuzzy tools to correctly > interpret > compared to having a language (extension) tag like <owlx: > LeftShoulderFunction> which again crisp tools will have to skip > while fuzzy > to parse and interpret. > > I think that only after thorough thinking, practical investigations > and > practice we could learn which one is better. At this point I see > just some > pros and cons for both of them. > > Cheers, > -gstoil > > >> Cheers, >> >> -Umberto Straccia >> >> >> %----Straccia's previous email >> >> The OWL ontology can be used to describe different types of >> uncertainty, but can also to used to describe HOW uncertain >> information is represented in OWL (without any language extension) or >> RDF/RDFS. >> >> To be explicit, suppose I would like to express the concept of YOUNG >> using an explicit fuzzy membership function such as a left-shoulder >> function with parameter a=20, b=30 (below 20 someone is young to >> degree 1, after 30 he is young to degree 0, in between we use linear >> interpolation). >> >> Now, we have two options >> >> a) either we suggest an extension to OWL (more precisely OWL-DL) or >> RDF/RDFS to explicitly accommodate such functions (i.e. we propose a >> language construct for that) >> b) or we develop an OWL-DL ontology (or RDF/RDFS ontology) describing >> HOW to represent such information into the current standardized OWL- >> DL language. In our specific case, we may say that >> >> LeftShoulderFunction ISA FuzzyMembershipFunction >> WITH HasParameterA of type Float >> WITH HasParameterB of type Float >> >> >> Then we can represent Young with >> >> Young ISA VagueConcept >> WITH HasFuzzyMembershipFunction of type LeftShoulderFunction WITH >> HasParameterA = 20 >> WITH > HasParameterB >> = 30 >> >> It is then up to a parser to correctly interpret the statements and >> then to load them into an underlying reasoning system. >> >> In this way our objective is (using the use cases) to develop an >> ontology, which describes HOW different notions of imperfect >> knowledge is represented in OWL-DL (RDF/RDFS). >> >> -Umberto. >> >> >> %---- >> >> >> On Jul 18, 2007, at 7:30 PM, Giorgos Stoilos wrote: >> >>> >>> Hi Ken, >>> >>> This sounds reasonable enough. But let me also stress another issue. >>> >>> A proposed extension should be as minimal as possible in order to >>> enjoy >>> acceptance by the non-uncertainty community and persuade people >>> that it >>> could be adopted in their tools with a minimal effort. So also >>> replying to >>> Peter's mail, I do not agree with extensions like >>> owl_ursw:usualy_oneOf, >>> owl_ursw:often_oneOf, owl_ursw:probably_subClassOf, which to me do >>> not seem >>> minimal. >>> >>> -gstoil >>> >>>> -----Original Message----- >>>> From: public-xg-urw3-request@w3.org [mailto:public-xg-urw3- >>>> request@w3.org] >>>> On Behalf Of Ken Laskey >>>> Sent: Wednesday, July 18, 2007 3:02 PM >>>> To: Peter.Vojtas@mff.cuni.cz >>>> Cc: public-xg-urw3@w3.org >>>> Subject: [URW3 public] OWL extensions [was Re: [URW3] ... three >>>> questions >>>> based on the last telecon] >>>> >>>> >>>> Let me make a suggestion of a minimum criteria for adding >>>> extensions: >>>> >>>> an extension can be proposed only if you can show its use in the >>>> context of an already discussed use case. >>>> >>>> This is motivated by several thoughts: >>>> 1. we will have solid examples of the extension; >>>> 2. we can more easily compare value of the extension against others >>>> proposed; >>>> 3. we move forward our analysis of the use cases; >>>> 4. if proposed extensions can be demonstrated only in the >>>> context of >>>> use cases already discussed, some of you will have the >>>> motivation to >>>> volunteer for leading the discussion at future telecons :-) >>>> >>>> Do we have agreement on this proposal? >>>> >>>> Ken >>>> >>>> P.S. per survey results, I will be setting up telecons for >>>> August 1, >>>> August 22, September 5, and September 19. >>>> >>>> On Jul 18, 2007, at 7:20 AM, Peter Vojtas wrote: >>>> >>>>> >>>>> Colleagues, >>>>> let me note that this wonderfull discussion has >>>>> started by >>>>> questions about the nature of "sentence" and "proposition", and I >>>>> have >>>>> added a word used by W3C documents "statement" and as an example a >>>>> triple. Of course a RIF rule can be also a subject to attachment >>>>> of an >>>>> uncertainty. >>>>> I think this can be satisfactory solved by using current W3C >>>>> standards and interpretation of them. >>>>> >>>>> Now the problem has shifted a little bit further, to >>>>> ontology. My >>>>> impression is that we need to have some (easy) examples in the >>>>> begining >>>>> (Ken already assigned some sentences in his use case by >>>>> uncertainty >>>>> type >>>>> and nature). >>>>> I like Mitch's ontology and so far only few extensions were >>>>> sugested - to have properties includesSentence, isaboutSentence >>>>> and a >>>>> new sort of uncertainty models namely Similarity (maybe some >>>>> other will >>>>> appear later - what are our criteria to enter new elements to >>>>> ontology). >>>>> The reification discussion was only an example from my part, >>>>> and can >>>>> be soved >>>>> by Uncertainty has_derivation objective/subjective. >>>>> >>>>> I have also an idea and would like to ask ou for opinion. >>>>> Most of >>>>> Ontological knowledge is described by expressions about being an >>>>> element >>>>> and being a subset (equal to), e.g. >>>>> >>>>> owl:oneOf, rdf:type, ... rdfs:subClassOf, ... >>>>> >>>>> what do you think about extensions like >>>>> >>>>> owl_ursw:usualy_oneOf, owl_ursw:often_oneOf, >>>>> owl_ursw:probably_subClassOf >>>>> >>>>> or we are just going to assign uncertainty to the statement >>>>> A rdf:type B, C rdfs:subClassOf D, ... >>>>> >>>>> I agree that sentences can be structured by logical connectives, >>>>> and I >>>>> would be here very flexible and allow also fuzzy aggregation >>>>> operators. >>>>> >>>>> On the one side we are not going to specify syntax but we have to >>>>> show >>>>> current standards are not necessary (of course not because of the >>>>> syntax of current standards - using W3C syntax we have in mind >>>>> that >>>>> their semantics does not suffice) >>>>> >>>>> Greetings Peter >>>>> >>>>> >>>>> >>>> ------------------------------------------------------------------- >>>> -- >>>> --- >>>> ----- >>>> Ken Laskey >>>> MITRE Corporation, M/S H305 phone: 703-983-7934 >>>> 7151 Colshire Drive fax: 703-983-1379 >>>> McLean VA 22102-7508 >>> >>> >>> > > > -- ______________________________________________________________________g5 m Artificial Intelligence Group || Phone: +44 117 928 8200 Department of Engineering Maths || Fax: +44 117 925 1154 University of Bristol || trevor.martin@bris.ac.uk BS8 1TR, UK || www.enm.bris.ac.uk/ai/martin/ ______________________________________________________________________g5 m Intelligence is Fuzzy : www.fuzzieee2007.org
Received on Thursday, 19 July 2007 19:16:02 UTC