> Can PurpleWiki be modified to be an ontology editing tool? My guess
> is yes, but the truth is that I'm not sure. Or, more accurately, I'm
> not sure if it's worth it.
>
> I think that Wikis already come close to mapping to a Topic Map model,
> as I explain at:
>
> http://www.eekim.com/cgi-bin/wiki.pl?WikisAsTopicMaps (01)
I think it would be throwing the baby out with the bathwater to make the
Wiki more complex - Nooron's a great idea, but I reckon too much is exposed
here : http://www.nooron.org/know/computing_systems_ontology (02)
Coming at it from an RDF/OWL point of view, I'll try by saying we want Wiki
pages to describe the People and PurplePeople classes, I think that could
look something like this: (03)
[[
ClassPeople
------------
ClassPurplePeople (04)
CategoryOntology
]] (05)
[[
ClassPurplePeople
------------
ClassPeople (06)
EugeneEricKim
ChrisDent (07)
CategoryOntology
]] (08)
The page title gives the class name, and asserts it's a class. The two
people included would be read as instances of the class (no prefix). (09)
I'm not sure about CategoryOntology, but something would be needed to say
'give this page special treatment' to any processor. (010)
But the problem is exactly as you describe: the links aren't typed. The
ClassPeople page might (should?) have a link back saying ClassPurplePeople -
which way would the hierachy run? (011)
So is here an easy way of typing links? The best I could come up with went
like this : (012)
A regular (non-ontological) link looks like [http://example.org title]
this could be extended to read [http://example.org title prefix:property] (013)
I'm not sure exactly what to do about CamelCase links, but what about
[WikiPageTitle prefix:property] (014)
where prefix is the namespace prefix and property is the specific
relationship. (015)
So in the example above the link in ClassPeople would read (016)
[ClassPurplePeople rdfs:subClass] (017)
Somewhere there'd need to be a map between namespace prefixes and the full
URIs, but that should be straightforward. (018)
When pages get created/modified, the parser sends messages to an RDF API to
create the appropriate statements. A link somewhere will allow it to be
viewed as RDF/XML, n3 or whatever. Alternately, it could be done lazily -
only when an 'export' link is clicked does the RDF extractor go to work, but
I think there is an advantage in realtime creation of the RDF model. (019)
I've already made a start on coding some of this stuff in my own
Wiki-hacking (I want a Wiki interface to IdeaGraph) but so far have only got
as far as creating the really obvious metadata for page creation (dates etc,
using the Jena API). Another design choice has meant I've got quite a bit of
other parser-related work to do (I want the Wiki pages persisted as XHTML,
WikiText only used for editing). (020)
Cheers,
Danny. (021)
_________________________________________________________________
Message Archives: http://dao.cim3.net/forum/dao-forum/
Subscribe/Unsubscribe/Config: http://dao.cim3.net/mailman/listinfo/dao-forum/
Shared Files: http://dao.cim3.net/file/
Community Wiki: http://dao.cim3.net/wiki/
Community Portal: http://dao.cim3.net/
To Post: mailto:dao-forum@dao.cim3.net (022)
|