Re: Chomsky AIML set

Hi,

Thank you very much for this interesting example.  I had a few stumbles
getting it into AIMLpad, but it is working now.  Here are some of my
statistics:

After removing the categories without a <template> tag and the ones that use
the Pandorabot's version of <learn> I came up with 178,175 categories.
That's a lot of AIML!

I ran the shadow report of AIMLpad and discovered 39,900 categories were
covered by other categories and would never be activated.  With a little
reasoning I guessed this may be because there are duplicates.

When the duplicates were combined into single categories, the count was
155,600 categories - still a very impressive number!

Then only 17,319 categories were shadowed. This makes a total of 138,281
active categories in the set.  That's nearly 100,000 more categories than
the AAA set - wow.

There are 20 categories which recurse into themselves.  If triggered, these
categories will use up the processing stack.  Here are some examples:

Pattern: _ EVER HAD A HUG  Template: <srai>HAVE YOU EVER HAD A HUG</srai>
Pattern: _ EVEREST         Template: <srai>MOUNT EVEREST</srai>
Pattern: HOW MUCH IS TEN * Template: <srai>HOW MUCH IS TEN <star/>
</srai>...
Pattern: WHO IS KAISER SOSAY

A few of those self-recursive categories would have been covered by a
category without a template, but I removed those to pass AIMLpad's loading
edits.

There are a few other minor edits I'd like to perform when I get more time
such as templates which use <star/> when there is no star in the pattern,
etc.

Normally loading the AAA set into the graphmaster index takes a second or
two.  Indexing the Chomsky set takes almost twenty seconds on my 1.6
gigahertz processor.  After that one-time startup, the responses are fast
again.  I've had a few conversations so far. Chomsky thinks I'm boring.

Again, thanks for the impressive AIML set.  Contact me if you want the
scrubbed up version or the list of shadowed categories.

Cheers,
  - Gary Dubuque, researching NLG to resolve the multiple sentence input
problems.

-----Original Message-----
From: alicebot-general-bounces@...
[mailto:alicebot-general-bounces@...] On Behalf Of
plafferty@...
Sent: Tuesday, August 12, 2008 1:27 AM
To: alicebot-general@...
Subject: [alicebot-general] Chomsky AIML set

The entire AIML set for the chatbot <a
href="http://tinyurl.com/ydxlef">Chomsky</a> (over 180,000 categories) is
now available for download at <a
href="http://tinyurl.com/5dxd25/chomskyAIML.zip">http://tinyurl.com/5dxd25/c
homskyAIML.zip</a>.

plafferty


-----------------------------------------
This email was sent from supanet.com.
http://www.supanet.com/

_______________________________________________
This is the alicebot-general mailing list
Reply to alicebot-general@...
Unsubscribe and change preferences at
http://list.alicebot.org/mailman/listinfo/alicebot-general
Learn netiquette at http://www.dtcc.edu/cs/rfc1855.html
Learn to read at http://www.literacy.org/


_______________________________________________
This is the alicebot-general mailing list
Reply to alicebot-general@...
Unsubscribe and change preferences at http://list.alicebot.org/mailman/listinfo/alicebot-general
Learn netiquette at http://www.dtcc.edu/cs/rfc1855.html
Learn to read at http://www.literacy.org/