pcplantdb

[Chad Knepp <pyg@galatea.org>]

John Schinnerer writes:
 > Aloha,
 > 
 > > Wow, I'm surprised and a little hurt by this evaluation of the current
 > > version (0.3.1) of PIW on <http://dev.permaculture.info/>.  While your
 > > evaluation is highly accurate of the current live version I think very
 > > significant progress has been made on dev.
 > 
 > Some time back you said you were going to go ahead and move dev to live 
 > so I assumed it already was.  However I see now it's not.
 > This is one of the problems with such an unorganized 'project'.
 > 
 > Meanwhile if I enter
 > 
 > shade loving legume
 > 
 > in the search box the first item returned is a bamboo (which although it 
 > apparently is OK in partial shade does not appear to be "shade loving" 
 > in any major way...).
 > Nothing on the first page is a legume.
 > 
 > If I enter
 > 
 > legume shade loving
 > 
 > nothing on the first page is a legume.
 > 
 > If I simply enter
 > 
 > legume
 > 
 > nothing on the first page is a leguminosae
 > 
 > Meanwhile if I enter
 > 
 > shade loving legume
 > 
 > in google I get about five immediately relevant hits on the first page, 
 > actually links to pages with information about some specific 
 > shade-loving legume plant or plants.  One or two even have a hint of 
 > companion plant/guild allusions in the info provided.

This is partially because the controlled vocabulary doesn't know what
you mean by 'legume'.  Do you mean all members of the family
Leguminosae, a nitrogen fixing deciduous tree/shrub, both, or
different things than that.  Currently this means that you have to add
something to the Culture dictionary (Eden/server/Vocabulary.py) like:
    
    'legume': "nitrogen_fixer = \"Y\"",

It already recognizes the following as well:

    'nitrogen': "nitrogen_fixer = \"Y\"",
    'fixes': "nitrogen_fixer = \"Y\"",
    'fixer': "nitrogen_fixer = \"Y\"",

Hmmm, looks like we are working on this again;-)  Seriously this is
the sort of thing that we should be discussing technically and
something I would like to be issued on.  Long term I think tagging
will replace controlled vocabulary so there will be a mechanism to
allow users to 'teach' the engine new words.

 > Nothing personal Chad.  It's just that with my hard-ass testing hat on, 
 > the above tells me that all else being equal I'm going to use google to 
 > find plant information.
 > 
 > > ....That said I think that much of the framework for gathering
 > > this data is in place, and as it is today PIW is EXTREMELY useful for
 > > figuring out what plants might work well with others,
 > 
 > How? Please give some examples.

The advanced search <http://dev.permaculture.info/advanced.html>.  The
simple search is really not for serious use unless we put a lot of
effort into developing the vocabulary.  It's great for looking up
'eleagnus' and having it give you what you meant but it's too young
for serious use.  The advanced search gives you direct access to the
entire (almost) dataset.  Hmmm... just found a major bug: suitable
locations doesn't seem to work from the advanced form.

Here's how I surf for plants:  I start off with some parameters like
climate zone (5b for me, so sometimes I pick 6 if I know it's going to
be in a warm microclimate).  I have pretty clayey soil everywhere so I
might add tolerates heavy clay or prefers clay soil.  Thinking more
about the microclimate I might add something about the moisture
level.  Is this a shady location?... or will it be?  If I'm looking
for a pioneer I pick fast growing and fixes nitrogen.  I use
tree/shrub/climber/grass to get approximately sized things.  Wow,
alders (Alnus sp.), I don't know anything about them....

 > Maybe the best thing to do is to announce it widely and open it up to 
 > all comers to play with.
 > That would get some quantity of real-world feedback as to its usefulness.

I agree, seems like folks are banging on it again.  This is part of my
desire to announce that it *open for testing* to the permaculture list.

 > With me saying it's pretty useless and Chad saying it's EXTREMELY 
 > useful, I'd say we need a lot more feedback.
 > So I reverse my previous postion that it shouldn't be public yet.
 > Put it out there and let's see what happens.

I'm working on it, but don't expect it today.


Sean Maley writes:
 > --- Chad Knepp <pyg@galatea.org> wrote:
 > > version (0.3.1) of PIW on <http://dev.permaculture.info/>.  While your
 > > evaluation is highly accurate of the current live version I think very
 > > significant progress has been made on dev.  Specifically the advanced
 > > search addresses most of your concerns
 > > <http://dev.permaculture.info/advanced.html>.  Many of the features
 > > that Ad Vlems listed are available as well. 
 > 
 > I put either cordage or fiber in the uses box, but still get the
 > entire dataset back.  Does this mean that everything in Eden is
 > useful for cordage and fiber?  I get more easily accessible
 > information from Google, which often leads me to PFAF.  Why should
 > I use PIW, instead?  This includes dev.

The cordage bit sounds like a bug.  That's two today already...

The accessibility is what we are working on... Anyway, hopefully PIW
will eventually serve as a general repository of permaculture
information... an open system that generates it's content from it's
users (permaculturalists).  Ex: My friend noticed the other day that
Luffa isn't in there.  You could create an account and add the
botanical name (correctly speeled please) as well as whatever you
happen to know about luffas.  Next week a luffa expert could look up
the luffa entry and see that you really don't know that much about
luffas and write about her own experience and research.  Other users
interested in luffas could vastly benefit from the comparative
analysis and you may even change your entry based on further
experience in an effort to compete in relevance with the luffa
expert. 

 > I don't like using Python.  I've been programming since the Z-80
 > days of CP/M and no longer have any favorite languages/IDE (C was
 > my long time favorite and Perl was great in a pinch).  If I want to
 > put nails in, I use a hammer.  I get a screwdriver for screws.
 > Currently, it is the switching gears issue to take up Python.  I
 > happen to have a day job using Sybase, Perl, and PHP (I'm the DBA,
 > but they give me extra work to help me look busy).  I'm putting my
 > foot down to avoid learning yet another tool.  Yes, I've read
 > through your Python code and you have very clean programming
 > practices.
 >
 > The database I saw, eden, was normalized minimally, though just a
 > plant repository.  I'm hearing OLAP concepts in the specification,
 > but little discussion on measures (dynamic data like soil
 > measurements and harvest quantity/quality) and dimensions (static
 > data like species and uses).  I can see that no measures exist in
 > the current DDL and little dimension, aside from PFAF plant data.
 > Some of the dimensional subjects could be mined from PFAF, but so
 > much more would be needed for the full dataset of guild dynamics to
 > be accomplished.  From past discussion, I'm sure I have a larger
 > scope in mind than others.  Perhaps I just don't see how much can
 > be done with just plants and no measures for PIW users to collect.
 > Are there even heuristics that I missed?

I am not a database expert and would appreciate more assistance here
with the provisions that the transition retains the PFAF dataset and
does not consume a huge amount of time.   

I think the dataset still needs a lot of work and there are a lot of
more types of information that we intend to add.  There are many
things missing...

Don't forget solstice,
Chad

-- 
Chad Knepp
python -c 'import base64;print base64.decodestring("cHlnQGdhbGF0ZWEub3Jn")'