pcplantdb

[Bear Kaufmann <bear@ursine-design.com>]

Hi all,

I'm back in town. Good to see some discussion.
I'm wondering if anyone got a chance to look at the RDF ontology I put 
up yet.
At:
http://www.permaculture.info/doc/PIWOntDoc/
Human readable
http://www.permaculture.info/doc/piwontv0.1.rdf-xml.owl
Machine readable

I do like the freedom that a wiki provides. But I see some arguments for 
adopting RDF descriptions at some point namely:
RDF is machine readable.

Why is that valuable?
Well, if a plant description is basically basically a page of text 
(ignoring the back end), that description may be very thorough and 
descriptive, and useful to a reader, but it leaves the plant as a rather 
separate entity, isolated from the rest.

In the ontology I was working on, at the organism level, I was wanting 
to have set of properties that at their root were: produces, consumes, 
traits, requirements, growing areas, etc. There might be subproperties 
which provide subtler distinctions, like something being the main agent 
of production (nest building), versus say producing EdibleFruit.

Having this knowledge in a machine readable form allows us to create 
something that can at least guide in the creation of food webs. For this 
design we want EdibleFruit, FixedNitrogen, Firewood, and 
WildlifeHabitat. Given a site's climate, soil, and location 
[EnvironmentConditions] (which limits the usable set based on organism 
Requirements), we can then find what groupings would provide those, 
perhaps gaining group score if they have other mutually beneficial 
connections, or lowering it for negative interactions.

This provides a richer potential.
But I don't think any of these strategies are mutually exclusive. In 
fact Rich's description of WikiMedia syntax provides the best of both 
worlds. If people can enter text using some similar markup 
[produces:Nectar] (the page is assumed to be the first part of the 
triple Salvia_egalans:produces:Nectar), those tags can be added to an 
RDF store, and the syntax can be defined and checked against an RDF 
Schema (probably the next level up in descriptiveness: OWL). Suggestions 
could be provided to the user regarding the available tags (which would 
show what's available in that context, ie: Plant, SoilType, Animal, 
Construction, etc).
This provides a nice descriptive page, but gives access to something 
that can model the network of connections/requirements, etc. Without 
some form of data that's readable in that way, we couldn't accomplish 
the visualizations that Stephanie was interested in having.

And all this could also be done in a more atomized way as Chad envisions.

Glad your OK Chad. Youch.

Cheers to all,
Bear