[pcplantdb] Raining Delectable Bruschettas

Bear K bear at ursine-design.com
Wed Feb 9 03:45:40 EST 2005


Hi folks,

Hmm, interesting subject.
Good to see a nice increase of traffic and discussion on the schema 
issues. I promise to avoid sending Phase 3 visions of front end system 
to the list for a while. :-)

> A quick something I've been thinking about while wading through 
> posts...
I'll reply while wading through the torrent of posts.

> Rich wrote, “It does open up some interesting technical Q's, on how to
> merge two databases. I suspect this is the sort of thing we will have 
> to
> deal with in the future, so now would be a great time to explore the
> problems of integrations.”
>
> Sean wrote, “.. spelling error, scraping. Basically, we do what google
> does. We scan the web pages from other plant database web sites and
> archive the information with our own data.”
>
> And then Lawrence asked, “Would it be possible for PIW to have/host
> multiple databases, each with different structural needs?…Does this 
> mean
> that someone using PIW could request info that we don't have but
> USDA does…”
>
> And then Bear said, let there be RDB.  hey Bear, isn’t all this what 
> RDB
> is used for?
Slightly scrambled acronym there (gotta love 'em), it's RDF.
Not everything above.
Actually more about this thread:
>>> Benifical relationships
[snip]
>> May I suggest that this simply be "Relationships", which can then 
>> include antipathetic/allelpathic information.
>> For me this relationship information (+, o and -) is usually the 
>> hardest to come by, and most crucial.
> Indeed this does seem like a good way to simplify the design.
It's really about defining a schema for relationships. It uses so 
called "triples".
Subject:predicate:Object

Even data fields are considered a relationship.
User2300:firstName:Bear

Ladybug:eats:Aphid
If a relationship is bi-directional, we can assume Aphid:isEatenBy:Aphid

For a primer see here: 
http://xulplanet.com/tutorials/mozsdk/rdfstart.php

I imagine that there probably are a reasonable number of DB's with 
eats/isEasten type relationships being built by ecologists already.

> Rich wrote, “It does open up some interesting technical Q's, on how to
> merge two databases. I suspect this is the sort of thing we will have 
> to
> deal with in the future, so now would be a great time to explore the
> problems of integrations.”
One thing with RDF is you can overlap schemas. So there's already a 
definition for personal information: http://xmlns.com/foaf/0.1/
There are other projects using RDF for biology and botanical purposes. 
If there is something that's an agreed upon schema for a plant naming 
system, we might decide to use that spec for that. If other projects 
use that too, we'd start to speek the same language, and could compare 
notes (merge/xreference) more easily on things each others domains did 
or didn't cover.

A man Steph and I meet at the Planetwork meeting said he had interest 
in the project, and has experience in developing schemas with RDF. He's 
used Protege to develop with. Find it here: 
http://protege.stanford.edu/ It's something that I looked at before he 
mentioned it. I believe it has a collaborate checkin/checkout system to 
develop with, and editing the schema with data using it doesn't corrupt 
the info. We might want to talk with him some more. Drawing up this XML 
DTD/Schema/RDF/etc agnostic description is very good, we could send to 
to him and see what he thinks.

OK, long day. Going to stop there. :-)

Cheers,
Bear



More information about the pcplantdb mailing list