Computer Programming for Everybody, a Newbie Project

Ron Stephens rdsteph at earthlink.net
Wed Sep 17 02:49:29 CEST 2003


I am planning on going forward in several ways, and I would appreciate
any feedback on which directions seem most worthwhile.

1. I plan to do a full GUI version of the entire program using
PythonCard. I also, eventually plan to do an alternative GUI version
(or two), certainly one using Jython and SWING (so I can make a
self-installing GUI version for my Sharp Zaurus). I have also loaded
PyQt on my Zaurus, but I find Jython and SWING to be easier for me to
learn and use. Maybe a PyQt version will follow, and I would also like
to do a Tkinter version for comparison purposes. I may use EasyGui to
create a simple Tkinter version, and I will probably do an AnyGui
version. Another Sharp Zaurus user has reported getting an AnyGui
program working fine on his Zaurus (AnyGui uses the QT tool kit on the
Zaurus).

2. I plan to use NLQ, a Python natural language processor module, to
allow Merlin to choose appropriate possible options, or answers, to
users' questions, and then applying the standard askMerlin web
scraping algorithm to decide which possible answer is best. I have
preliminary versions of this working, but not yet posted, because it
doesn't choose good enough options yet, but I think I have figured out
ways to improve it. Next, I plan to make a more interactive and
versatle version of general decision analysis and askmerlin that will
flexibly adapt to a user's entries, answering a user's random question
most optimally, whether the user just asks a question and gives no
options, criteria, or weights, or whether the user gives any given
combination of the above. In other words, instead of making the user
declare upfront how much information she has to offer, it might seem
more fluid to the user to use one generic module that answers
questions regardless of how much information is given, although of
course the quality of the program's answers will tend to be better
when the user gives more information up front.

3. Concerning the web scraper algorithm used in askmerlin, I am
thinking about some ideas to do a version that will automatically
track trends on the web over time, graphing the output, in response to
specific user questions.

4. I want to do an online version the whole program, by making a
server-side version of askMerlin. Probably starting with a simple CGI
version, the idea will be to post the program on my website, and allow
a user to interact over the web; but this will also lead to the
storage of the sessions, allowing a database to be built up which
tracks user inputs, especially on scoring the various options for each
criteria. Picture this with respect to the election.py module. For a
given election, say the 2004 USA Presidential election, once we are
down to 2 or 3 candidates for the general election next November. As
each user enters their scores for each candidate on each criteria, not
only will they get an immediate response to the over all score of each
candidate based on their responses, but more significantly the
database can tally all users scores and average them, and plot the
results over time. I suspect that, for a website with enough users,
the data so generated could be valuable. My site currently gets about
10,000 unique visitors per month. If a site could get 1,000 users of
election.py (online version) per month, this might usefully supplement
the official Zogby and Gallup type "polls". No, the users would not be
scientifically sampled and randomly selected, and probably a site
would need to have, say, 100,000 unique visitors per month in order to
get 1000 to actually do the election.py online, in order to get useful
results. But still it is interesting to me, and it shouldn't be that
hard to make such an online server side version of election.py, or so
I figure. ;-))))

5. I will no doubt add at least a few new modules on new topic areas
to Decision Analysis. Of course, my biggest hope is that others will
do such additions and share them with me ;-)))


Any other ideas would be appreciated. Meanwhile, I would be interested
if anyone would offer thoughts on which of the 5 directions for future
activity sound most interesting.

Ron Stephens
http://www.awaretek.com/plf.html




More information about the Python-list mailing list