- From: Golda Velez <gv@btucson.com>
- Date: Sun, 3 Feb 2008 23:40:40 -0700
- To: Tim Berners-Lee <timbl@w3.org>
- Cc: semantic-web@w3.org
Adrian, Tim, Paola, Thanks very much for the replies! Starting here... On Sunday 03 February 2008 9:38, Tim Berners-Lee wrote: > Golda, > > The data web is most appropriate for data, things which can be > processed and re-used in various ways, as they have fairly well-define > meanings. Your example has a bunch of "maybe indicates" sort of > things. Yes, but...in most scientific fields, there is disagreement about the definitions of data, and in politics (a game, but 'the only game for grownups') even more so. In many cases heated disagreement - even about things like the definition of a nation, the extent of a promoter region, the nature of 'green' technology. I would say that if you restrict the semantic web to cover items with no controversy over definition, you either stay within autocratic corporate structures or rather boring, toy-like data. Tim, even your original 2001 paper has what I would consider a flaw in this regard, where you give confidence to a blackberry-type device to pick a medical specialist for your mother, based on convenience and insurance. I'm afraid I'd want to see opinions of other patients, doctors, and other non-authoritative data before choosing a doctor for my mom...you never know. So, I'd like to have a data web in which I can process and re-use even non-authoritative data. The disagreement in particular is what I'd like a computer's help quantifying and voting on - but in a more precise way than simple 'link popularity' of entire website URIs aka google. > Given that disclaimer, > here is one way of going about what you want. [snip] > > --------------------- > > "The Az Sonora Desert Museum serves shade grown-coffee, which supports > > ecological diversity as per Win-Win Ecology by Mike Rosenzweig." > > --------------------- Thanks for the help! I'm a raw beginner at N3 and RDF/XML, and didn't want to show off my ignorance. I'm afraid though that we didn't catch it all here; I'm the one who saw the 'shade grown coffee' sign at the Desert Museum, and I've read Mike's book, so now I'm posting this assertion putting the two together. Mike didn't know the Desert Museum was serving this coffee, he just explained how important it is in his book. So, I've modified it a bit, would this make sense? @prefix dc: <http://purl.org/dc/elements/1.1/> . @prefix foaf: <http://xmlns.com/foaf/0.1/>. @prefix ex: <ont#>. # @@ write ontology :wwe dc:title "Win-Win Ecology"; dc:creator [ foaf:name "Mike Rosenzweig" ]. <http://www4.wiwiss.fu-berlin.de/bookmashup/books/0195156048> = :wwe :asdm foaf:name " Az Sonora Desert Museum". :sgc foaf:name "Shade-Grown Coffee". :ecdiv foaf:name "Ecological Diversity". # though it seems to me a pointer to a place in a hierarchy like dmoz would be # more useful than a foaf:name, maybe taxo: something? :asdm ex:serves :sgc {:sgc ex:supports :ecdiv } ex:accordingTo :wwe #------ Will get back to the ruleset in a moment, but now suppose a representative comes around from a big full-sun coffee plantation and wishes to disagree with my assertion. Can he refer to it, or should he just make his own separate assertion about :sgc and :ecdiv? Can we keep track of who I am, who made the first assertion, and who he is? Am I badly misusing RDF? --------- > #Also, we believe > > :sgc a ex:ProEcologyProduct. # we just have to change this to make me a trusted source, in my context, and # the ruleset still works. I"m not sure the best way to do that here. > :wwe a ex:TrustedSource. > > # Then I would personally use a rule > > @forAll :business, :product, :source. > { :source a ex:TrustedSource. > { :business ex:serves :product } ex:accordingTo :source. > :product a ex:ProEcologyProduct. > } => {:business a ex:FavoredBusiness}. > > # Running these rules with cwm --think will produce: > # :asdm a ex:FavoredBusiness I like very much the abilities of cwm --think, but I hope it has some fuzziness or voting capabilities? After all, :asdm is a big place, and perhaps someone else might notice that one of the snake's cages is too small, and put it a note about an animal not being cared for well, which would result in a non-favored business status - so keeping track somehow of the votes, trust levels and reasons would be valuable too? Doesn't mean it has to be in the tool now, but is RDF amenable to storing this kind of fuzziness? --------- Well, I am probably out of line here, as I've been reading a lot more than I've been coding RDF. Any suggestions as to the best use cases today for web developers to apply semantic web technologies are very welcome! As are 'didn't you read the FAQ' responses - just tell me which one! Sincere thanks for taking the time to consider all this! --Golda [no further new comments below] > > @prefix dc: <http://purl.org/dc/elements/1.1/> . > @prefix foaf: <http://xmlns.com/foaf/0.1/>. > @prefix ex: <ont#>. # @@ write ontology > > > > > :wwe dc:title "Win-Win Ecology"; dc:creator [ foaf:name "Mike > Rosenzweig" ]. > > #Read, wwe has a title ... and has a creator which has name Mike ...: > #(Use cwm to convert that into RDF/XML if that is easier to read fro > you) > > #I'd note that that book has a URI in the mashup, which we could have > used. > # Or we can just say > > <http://www4.wiwiss.fu-berlin.de/bookmashup/books/0195156048> > = :wwe . > > > #Now the museum. We'll need new ontology terms I'll put in ex: > namespace. > > :asdm foaf:name " Az Sonora Desert Museum". > :sgc foaf:name "Shade-Grown Coffee". > > { :asdm ex:serves :sgc } ex:accordingTo :wwe. > > # There are various "accordingTo" type verbs one can define > #---------------------------------- > > #Also, we believe > > :sgc a ex:ProEcologyProduct. > :wwe a ex:TrustedSource. > > # Then I would personally use a rule > > @forAll :business, :product, :source. > { :source a ex:TrustedSource. > { :business ex:serves :product } ex:accordingTo :source. > :product a ex:ProEcologyProduct. > } => {:business a ex:FavoredBusiness}. > > # Running these rules with cwm --think will produce: > # :asdm a ex:FavoredBusiness > > > > > I can use wikipedia or other authoritative URIs for the entities and > > concepts > > like the Desert Museum, shade-grown coffee, ecological diversity, > > and the > > book. > > Indeed > > > The question is which vocab's to use for the verbs - serves, > > supports, > > and 'as per' - > > The domain-specific verbs you may have to invent. > you might find ontologies of food. > Algorithm: Spend a limited amount of time looking for people who have > already defined terms in the area. then do your own for te missing > ones. Later, if you find more ontologies, build links between them. > > > and can I use reification or do I have to invent a tortured > > class that owns its own caveats? bagID would seem useful for this > > but its > > deprecated? > > > > Avoid reification. Do use nested graphs, as in N3. > > The file above: > > http://www.w3.org/2008/02/03-eco/a.n3 > > The same in rdf -- after inference, without the rules > $ cwm a.n3 --think --rdf --data > b.rdf > > Hope this helps > > > Tim BL > > -- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Golda Velez 520-440-1420 http://goldavelez.com what I do: Tucson Superblog http://btucson.com Search software http://webglimpse.net Web hosting http://iwhome.com "Help organize the world - index your own corner of the web!"
Received on Monday, 4 February 2008 06:37:06 UTC