Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: SUO: Try coding protocols




Adam,

At 11:34 04/09/01 -0700, vous avez écrit:
>Josiane,
>   Comments below:
>
>At 12:27 AM 8/31/2001 +0200, josiane caron wrote

>Adam, you know I want to modelize my protocols. I know it is not possible 
>to do it all things at a time.
>>At a first level, I know that it is necessary to code at propositional 
>>level, and to see how it will work there.
>>
>>Here I have an ultime question to ask you in case of the following example:
>>
>>10 - I take the pink disc I put it on the green disc
>>42  31  0 -----> 421 3 0
>>
>>(instance Person10 Human)
>>(instance Take10 Removing)
>>(instance Put10 Putting)
>>(origin Take10 Peg-B)
>>(patient Take10 Disk1)
>>(agent Take10 Person10)
>>(destination Put10 Peg-A)
>>(patient Put10 Disk1)
>>(agent Take10 Person10)
>>(after Put10 Take10)
>>
>>(inScopeOfInterest
>>   Person1
>>   (penetrates Peg-C Disk2))
>>
>>(inScopeOfInterest
>>   Take10
>>   (and
>>     (on Disk4 Peg-A)
>>     (on Disk2 Disk4)
>>     (on Disk3 Peg-B)
>>     (on Disk1 Disk3)))
>>
>>(inScopeOfInterest
>>   Put10
>>   (and
>>     (on Disk4 Peg-A)
>>     (on Disk2 Disk4)
>>     (on Disk1 Disk2)
>>     (on Disk3 Peg-B)))
>>
>>This would be the coding of the first level. But at the same time and at 
>>physical level disk1 is moved on Disk2 which is on pegA.
>>Te question is:
>>Is the coding sufficient to account of this fact ? (It depends of the 
>>status of 'inScopeOfInterest' in the whole coding)
>>Because here there are two physical entities : pegA and disk 2 on pegA. 
>>The boundary between physical entities and the others is not so clear. 
>>And I do not yet know what is really coded here.
>>
>>This was the basis of all my later questions (and then I was carried away 
>>in all sort of coding)
>
>I'm not sure what you mean here.  I think you may be addressing that there 
>are multiple levels of information that can be extracted from the verbal 
>protocols.  If that's the case, I may have been leading you astray since 
>I've been trying to provide you with logic statements that represent all 
>the information in the example, without the restriction of only addressing 
>the information contained on the surface of each individual statement.

Perhaps the ambiguity stands in the term 'level'. I want to mean different 
kinds of perspective taking upon information. Is it right for you like that ?
In fact I would like to have different parallel treatments with some 
possible inferential links at specific moments. But surely it is silly.

But you haven't answer my question:  Is the coding above sufficient to 
account of the fact that disk 1 is moved on disk 2 which is on peg-A
It is a matter of status of the 'inScopeOfInterest'. Does it work at the 
same time (the asame level) of the coding of the sentence ? and in this 
case we do not have to give some more precisions.


>>-----------------
>>
>>I have another question
>>
>>I like to introduce some properties of the physical state: for example 
>>the state is in the top right part of the triangle (the triangle is the 
>>set of every possible states). I think it is better to introduce it at 
>>another level than at the first level ?
>
>I'm not sure what you mean.  Could you provide a diagram of the triangle 
>you are referring to?


It is the triangle of all possible steps of the problem; I will send you it 
by file as soon as I will find it on my computer. (I have recently changed 
my computer and I have not yet found all my files)

>>----------------------------
>>At another level there is an important property - it is to give a 
>>differenciation in the way in which events are linked. For example to 
>>differenciatiate the two following utterances
>>
>>I take the pink disk I put it on pegB
>>I take the pink disk I put the pink disk on pegB
>>
>>In the first there is an anaphora
>>In the second there is only the repetition of 'the pik disk'
>>
>>This case is not frequent in Olivier protocols, but generally it occurs 
>>frequently. It is basic for the interpretation (differenciation between 
>>automatic and controlled processes)
>>
>>In Olivier we can find this occurrence where thereis anaphora between the 
>>two first events and not for the third
>>
>>9 - Then I take the pink yellow disc I put it on the column b
>>         (4 2) (1) (3)
>>
>>So that I can put the green disc on the column c #2
>>         (4) (1) ( 3 2 )
>>
>>This question is important because I was asking myself if we have to 
>>consider two events 'removing and putting' in the same proposition or if 
>>one have to consider a first propsition with the event 'removing' and 
>>another with the event'putting' and then at a second level to consider a 
>>link between these two propositions, with at a third level a 
>>differenciation between these links
>
>A the logical level both of the sentences above are the same.

I always thought it was.

>The challenge is for a NL system to generate the correct equivalent 
>statements from different NL sources.

I know, but it would very interesting now to follow the reverse way and to 
try to reconstruct some different source from this equivalent statements. 
It is this that I want to do in order to match these processes of 
reconstruction to the problem solving strategy.

>   I'm starting to think that if you're addressing this "pre-logical" 
> information, that logical statements may not be the right formalism for 
> these intermediate steps in translating from NL to logic.

This is not a logical thought. I asked you these questions because I am not 
familiar with these systems, and I like to ask silly questions rather than 
work without understand well.
But what do you mean exactly ?

Here there is the exams period now, and I have to go during next week-end. 
I will take again all this next tuesday when I will come back home;
Yours
Josiane

>>----------------------------
>>
>>
>>I think I have to wait for your answer before to express the other levels 
>>of coding
>>As soon as I have clear things in my head I will code trial1 and others 
>>and also some other protocols.
>>
>>Sincerely Yours
>>Josiane
>
>Adam Pease
>Teknowledge
>(650) 424-0500 x571