Skip to Content.
Sympa Menu

pcplantdb - Re: [pcplantdb] Synth, comments, etc

pcplantdb@lists.ibiblio.org

Subject: pcplantdb

List archive

Chronological Thread  
  • From: Bear K <bear@ursine-design.com>
  • To: Permaculture Plant Database <pcplantdb@lists.ibiblio.org>
  • Subject: Re: [pcplantdb] Synth, comments, etc
  • Date: Thu, 9 Dec 2004 15:09:02 -0800

That would require some Python Imaging Library handwork right? Assuming
you produced a still GIF image from it. Or at a later date get someone
to build a Flash based viewer using Actionscript to provide interactive
viewing (or TouchGraph, etc).

Or SVG <http://www.mozilla.org/projects/svg/>... the open source
equivilantof Flash.
Right, I'll take a look at that.

It seems that you could also just score your results from the search to
create a cumulative rank. Those with high marks on all three go first
on down, and just display it as a list (with the rank split into the
different columns, this system would theoretically work with any number
of variables, and doesn't require a special UI.

These are/should be front-end issues.

Yep, as I said, easy to do the searchs, hard to display the results
meaningfully.

<snip>

I am in favor of a more generalist approach
and personally think the best way to figure out a crop rotation or a
guild is with an observant human mind that is in touch with the best
info available (hopefully PIW). At the moment I just can't imagine a
computer system I would trust to really generate a guild or crop
rotation.
Making PIW the google of plant info. Greatly speeding research, not
just saying:
Based on X, Y, and Z, you should plant A tree, with B overstory, and C
vines.

But if say on the page for a single item view of plant info, it had
link to "find related plants".
Actually, a set of links to preformed queries "find functional plants
of this genus" (ie search current plant genus with rating of 3 or
over). "find plants with similar products", "find plants alleopathic to
this one".

Ok though, how do you represent these queries to the user? There are
more possible queries than one can display sanely so what queries
are/will be the most important ones and how should the end user
interact with these options in a browser based interface?

The searches I mentioned are just examples, but these are definite questions to work on, and luckily it should be easy to change these on the front end, unless they ask for some search functionality that isn't available.
I'd love to play around mocking an interface up, the next few weeks are very busy though.

I can embrace the concept of relationships because making connections
between information is an important concept, even if I don't currently
have a real clear idea of implement it. If oberservation has
determined following one annual with another is a beneficial
relationship this could be recorded as such and the interconnection
between the two plants could/should be displayed on both plants.

One interesting thought regarding this is:
Say you have a tree which requires X pollinator. Turns out Z plant
attracts Y pollinator. Thus you could write X <- is beneficial to -> Z.
What's lacking is that Y is the cause of the benefit. Z plant might not
grow under your conditions, so you need to search for other plants that
attract Y. It seems like it might be important to include Y somehow.
Perhaps the easiest right now would be: X is related [because of: Y] to
Z, where [because of Y] is just a text string. Later the schema could
grow to allow more info on Y to be added, and a script could run
through the field to build out the Y entries.

ARGH! My brain is burning! Seriously though, sometimes I wonder it
is possible to represent a web of relationships on computer that is
large enough to be meaningful. Every element has so many
possible relationships, is it possible to get them all and will it be
helpful if we can't. Not sure myself, but I'm committed to try.

I knew you'd *love* that one Chad. :-)

It seems like folks working on knowledge bases and such must have gone through things like this, though I seems there hasn't been a definitive (available) solution. I've been mulling over the idea of an application that could help people link objects and form PC networks. To put a UI in words: "Your looking at [chicken], in other networks chickens were [next to][compost pile], [above][pond], [inside][greenhouse],[in][orchard]...you could view systems where that specific relationship was used, etc. It all gets rather complex to develop in short order. Seems like objects (nouns) link through (verbs). Where do you draw the boundary on a system. Do you count the intestinal micro-fauna of a chicken as part of the system. If you do, can you lump the lower details together into a higher system level, and what kind of search function would do that. Items have constraints (only grows in Sunset Zones 1,2,5-7). I'm wondering if there is or are ways to represent this like a UML diagram or some such.

There was a semantic network project I was looking at once. The nodes consisted of words. A large number of people were given a random pairing of words, and asked to rank how similar or different they were. This information weighted the connections between the nodes, which when arranged using Elastic Interval Geometry (the rubber-band networks that TouchGraph[?] et al use), would lump into related clusters of words. I wonder the "Interaction Matrix of Two Species" (table 3.5 page 62 of the DM) could be used for a system like that. Would it just require a 1 to 1 table with columns for the rankings? You'd need another to store the individual rank, source, and comments that are averaged into the main table.

Also, I was just looking through www.flickr.com, a photo-blogging service, which has an API.
http://www.flickr.com/services/api/

Perhaps something like that could act as the interface between the back-end and front-end?
Then we'll have folks on Blogger with "My Favorite Plants" lists, and screen-savers which display plants which grow in your USDA Zone based on ZIP code. Plants with the most trackbacks.
Just kidding.

Let me know if all this talk hinders development.

Cheers,
Bear





Archive powered by MHonArc 2.6.24.

Top of Page