RE : Causation in SKOS?

Hi Marv,

There is nothing in SKOS for representing causality relations. I guess the good practice would be for someone to
1. introduce a custom property (or a set of such properties, as Alan as pointed out the matter is relatively complex) 
2. declare it/them as subproperty of skos:related to make them interpretable by general SKOS tools, and 
3. campaign for these custom properties to become a de facto standard for people interested in causality :-)

Note that your requirement is similar to the one of introducing in SKOS properties such as broaderGeneric or broaderPartitive which denote specific flavors of hierarchical links.
We did not include these ones in the standard: they are not of general use (even though they do occur quite often, indeed much more often than causality) and they might not be consensual (a big problem being some KOSs' representing part-of by hierarchical links and other by non-hierarchical links)

Cheers,

Antoine


-------- Message d'origine--------
De: public-esw-thes-request@w3.org de la part de Alan Ruttenberg
Date: dim. 16/11/2008 19:41
À: aunmguy@yahoo.com
Cc: public-esw-thes@w3.org
Objet : Re: Causation in SKOS?
 

If only it was so easy ;-)
http://plato.stanford.edu/search/searcher.py?query=causality

-Alan

On Sat, Nov 15, 2008 at 4:02 PM, Marvin Lugair <aunmguy@yahoo.com> wrote:
>
> Hello,
> Can someone tell me how and if causality is already represented in SKOS?
> I see that there are some categories in dbpedia that aim to do this, for example dbpedia contains the following triple:
>
> <http://dbpedia.org/resource/Western_Airlines_Flight_2605>
> <http://www.w3.org/2004/02/skos/core#subject> <http://dbpedia.org/resource/Category:Airliner_crashes_caused_by_pilot_error>
>
>
> But is there already a better way to represent causality using SKOS maybe? It would be nice to to be able to answer queries of the form: "what causes what"..
>
>
> I am pasting my related post below... Your imput is much appreciated!
> Marv
>
>
> ------------------
> Try the following SPARQL
>
> select ?p ?o
> where{
> <http://dbpedia.org/resource/The_Great_Train_Wreck_of_1856>
> ?p
> ?o
> }
>
> It will all the results the dbpedia has on that subject including:
>
> ?p = http://dbpedia.org/property/cause
> ?o = human error
>
> My question is why is the object a text string and not a URI?
> If you lookup a disease, say Malaria, you can find its cause in the abstract as text but you cannot ask the question what causes Malaria and have it answered in dbpedia. I know these are difficult to extract but has anyone thought about doing this?  Why is cause in the infobox of the train wreck but not in that of the disease? In Malaria the infobox points to a diseasedb reference (7728) can I follow that somehow to get the cause of malaria as a URI, and can that URI be used usefully with dbpedia?
>
>
>
> Another question is that why cant I find the train wreck triple when I do the following:
>    grep The_Great_Train_Wreck_of_1856 infobox_en.nt
>
> In fact I cant find it (the train wreck cause triple) in any of the files (I am looking at files i downloaded 2 months ago and querying dbpedia live but has this cause property been added to the infobox this recently?)
> ------
>
>
>
>
>
>
>
>
>
>