[alicebot-general] How interpreters handle multiple PATTERN <that> THAT <topic> TOPIC

[mehri]

Interesting discussion I had with a fellow by the name
of Ryan on how to handle duplicate "PATTERN <that>
THAT <topic> TOPIC"'s.  The AIML specification leaves
this open to interperters but I'd like to solicit
feeback from the community on what they commonly do
about the situation.

------------------------


Hi Ryan,

What I mean is that for example if you have a file
with:

<category>
<pattern>*</pattern>
<template>What you talking about willis</template>
</category>

<category>
<pattern>*</pattern>
<template>What you talking about nelly</template>
</category>

Two identical categories.  In graphmaster speak that's
two identical "PATTERN <that> THAT <topic> TOPIC"


The second one will override the first.  It's a
particularily nasty issue in which the specification
leaves open to interpreters to handle.  Real world is
that most AIML sets will end up having duplicate
"PATTERN <that> THAT <topic> TOPIC"'s.

So a person asks themselves when writing interpreters,
"What should I do?"  The options are:

* Overide the first with the second
* Never override
* Merge the two where it's a random toss on which one
is used.

> Someone might add a answer to I LOVE YOU
> when there already is one that says something like
> "Thanks, name, I like you a lot too."
> Then be confused by their answer only appearing half
> the time.  BUt they would be confused if sometimes
> theirs didnt happen at all, like when the first
> duplicate is selected over theirs.

Totally agree.  As a matter of fact, what I should be
doing and am adding to my dev-todo.txt list is going
to be a callback method on my CallBacks class that
alerts when two are clashing together.  That way users
of my AIML api will know when they have identical
"PATTERN <that> THAT <topic> TOPIC"'s.



__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com