[MUD-Dev] [Theory?] Parsing of knowledge to allow for more interactive NPCs

[Robert Zubek]

On Sun, 23 Dec 2001 holding99 at mindspring.com wrote:
> At 11:45 AM 12/22/2001 -0600, Robert Zubek wrote:

>>      "take the apple from the box and lock it"
>> vs.
>>      "take the apple from the box and eat it"

> An excellent example, and one I had not considered. In fact, I am
> tempted to disallow compound commands (two commands linked with
> the word "and") just to eliminate that possibility.

Well, this ambiguity comes not from the use of compound sentences,
but from pronoun binding. Consider someone saying, along the
examples you mentioned earlier:

  "examine the box"
  "take the apple from the box"
  "lock it"

Even though the sentences are simple, the last one remains
ambiguous.  On the other hand, a compound sentence such as

  "take the apple from Bubba and hit him with it"

is not so ambiguous, as the pronoun bindings are clear.

> While I wouldn't mind making some assumptions as to the player's
> intentions (such as eating an apple as opposed to a box), I am not
> sure how I will implement the assumption-making process. Perhaps
> flag each item with an "intended use".

As a possible way to implement this, you could just check which of
the items can fill the desired role in the given action, and then
bind the pronoun to whichever item fits best. Using the example of
"take the apple from the box and lock it", if an apple can't be
locked, it can't be the binding for the pronoun; but a box can so it
should get bound. In a sense, what this would be doing is treating
the pronoun as a wildcard, testing whether it would work with each
of the different values it can take, and picking the binding for the
noun that works.

This will not take care of all ambiguities, however, such as in:

  "take the apple from the box and drop it"

Since both the apple and the box should be droppable, the technique
won't disambiguate this sentence.

[ re ambiguity of interpreting prepositional phrases ]

> The second, and more insidious, is the use of the word "with" in
> terms of the knowledge chunks I posted before. It's actually not
> just "with" that causes a problem; "of" is equally problematic,

Yes! Several prepositions will give you similar problems. Some
examples off the top of my head, 'on', 'in', 'with' can apply not
just to the target, but also to the entire action:

    1. "hit (the man with the funny hat)" vs.
       "hit (the man) with (the funny hat)"
    2. "hit (the troll on the table)" vs. 
       "hit (the troll) on (the head)"
    3. "examine (the apple in the box)" vs.
       "put (the apple) in (the box)"

'Of' is differently problematic, in that it can serve both as a
quality of the object, and as a genitive descriptor:

    4. "examine the Sword of Doom" (the proper name) vs.
       "examine the ball of wax" (ie. ball made of wax) vs.
       "examine the armor of Bubba" (ie. Bubba's armor)

(also see: http://www.kanga.nu/archives/MUD-Dev-L/2001Q4/msg00342.php)

An interesting place to start might be trying to implement each of
these three cases separately (ie. ambiguities in pronoun binding, of
preposition scope, and of the 'of' preposition), because there seems
to be a good way to attack each one of these in separation, and see
how well these individual solutions work. This implementation would
likely then make it clearer if there is a general approach that
unifies them all...

Rob

--
Robert Zubek
rob at cs.northwestern.edu
http://www.cs.northwestern.edu/~rob


_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
https://www.kanga.nu/lists/listinfo/mud-dev