Re: assert vs instance-of and other questions

Dear Thomas and Hans,

Thank you very much for your helpful comments!

Best,

Kotaro

On 2006/01/13, at 2:35, Thomas Russ wrote:

>
> On Jan 12, 2006, at 3:01 AM, Kotaro FUNAKOSHI wrote:
>
>> Dear members,
>>
>> I have three questions about PowerLoom.  I will highly appreciate it
>> if somebody gives me answers.
>
> I will be glad to.
>
>>
>> (1) assert vs instance-of
>>
>> I found an asymmetricity between (A) asserting conceptual propositions using
>> command "assert" directly and (B) asserting conceptual propositions using
>> relation "instance-of".
>
> Yes.  This is a known problem.
> The fundamental issue has to do with PowerLoom rule indexing,
> which I'll cover in the answer to your second question.
>
>> Assume the following KB:
>> --------------------------------------------------------------------
>> (defmodule "PL-USER/TEST")
>> (in-module "PL-USER/TEST")
>> (clear-module "PL-USER/TEST")
>> (in-dialect :KIF)
>>
>> (defconcept animal)
>> (defconcept monkey (animal))
>> --------------------------------------------------------------------
>>
>> case (A):
>> When I use command "assert", everything is fine.
>>
>> |= (assert (george monkey))
>>
>> |P|(MONKEY GEORGE)
>>
>> |= (ask (instance-of george monkey))
>>
>> TRUE
>>
>> |= (ask (instance-of george animal))
>>
>> TRUE
>>
>> |= (ask (animal george))
>>
>> TRUE
>>
>> |= (ask (monkey george))
>>
>> TRUE
>>
>> case (B):
>> However if I use relation "instance-of", something goes wrong.
>>
>> |= (definstance george :instance-of monkey)
>>
>> |i|GEORGE
>>
>> |= (ask (instance-of george monkey))
>>
>> TRUE
>>
>> |= (ask (instance-of george animal))
>>
>> TRUE
>>
>> |= (ask (monkey george))
>>
>> UNKNOWN
>>
>> |= (ask (animal george))
>>
>> UNKNOWN
>>
>>
>> Why cannot PowerLoom answer that (monkey george) is TRUE when I use
>> "instance-of" even though it answers correctly against (ask (instance-of george animal))?
>>
>> I just wanted to use "instance-of" because I prefer frame-style descriptions
>> such as
>>     (definstance george
>>          :instance-of monkey
>>          :likes banana
>>          :at forest1
>>          :age 4).
>>
>> I can simply avoid this problem by using always command "assert" like
>>     (definstance george
>>          :likes banana
>>          :at forest1
>>          :age 4)
>>     (assert (monkey george)).  ( yet not so elegant :-( )
>>
>> Even avoidable, such a behavior of PowerLoom makes me feel ill at ease...
>> Is this an intended behavior or just a bug?
>
> This is just a bug.  I don't know when it will be fixed, though.
>
> Fortunately, there is a fairly easy alternate way of saying
> what you want that will allow you to use the frame-style
> descriptions.
>
> Any single place (unary) relation, which includes concepts,
> can be specified as if it were a boolean binary relation.
> In other words, instead of writing
>     :instance-of monkey
> you would write
>     :monkey true
> and this will cause PowerLoom to assert (monkey george).
> Your full example would look like
>
>    (definstance george
>          :monkey true
>          :likes banana
>          :at forest1
>          :age 4)
>
> This will then get you the inferences you want.
>
>> (2)  Warning: PowerLoom can't index the rule
>>
>> I tried to avoid the problem described above by using a rule. But I got a
>> warning message "Warning: PowerLoom can't index the rule"  and the rule
>> seems not to work.
>>
>> The rule is here:
>>  (defrule "instance-of TO assert"
>>    (forall (?x ?c)  (=> (and (concept ?c) (instance-of ?x ?c)) (?c ?x))))
>>
>> What's the meaning of the message and what's wrong with my rule?
>
> PowerLoom itself has a similar rule, but our sources also
> indicate that this is not properly indexed.
>
> What the warning means is that although the rule is a
> legal and correct rule, it has a part that is too complicated
> for PowerLoom's current machinery to index.  This has the
> unfortunate effect of rendering the rule useless, since without
> indexing, the PowerLoom reasoner can't find the rule when it
> would need to apply it.
>
> In your particular example, the problem is that the clause
> (?c ?x) doesn't give PowerLoom anything definite to use for
> indexing the rule.
>
>> (3)  Powers of PowerLoom
>>
>> This is a basic question about PowerLoom.  Is PowerLoom superior to or
>> equivalent to Loom in any respects?
>
>> Loom seems have more complex syntax (maybe syntax sugars) than PowerLoom.
>> But yet does PowerLoom have greater expressiveness than Loom?
>
> I'll try a quick and general answer.  If you want more details, ask.
>
> PowerLoom definitely has greater expressive power than Loom.
> That is because the language that can be used to describe
> concepts in Loom is more limited.  It is a description logic
> with certain additional constructs (:satisfies), but that is
> more limited a language than first order predicate calculus.
> Also, Loom has only limited ability to reason with :satisfies
> clauses, and in particular no way to classify concepts based
> on them.
>
> PowerLoom's input language is first order predicate calculus,
> which is more expressive.
>
>> Should everybody use PowerLoom instead of Loom? Or somebody with some
>> specific purposes had better use Loom rather than PowerLoom?
>
> It depends.  There are certain specific purposes where Loom would
> be better suited at the moment.
>
> Specifically, Loom would be better suited if you really care
> about concept classification.  PowerLoom has the ability to classify
> concepts and relations, but that form of reasoning has not yet
> received a lot of attention.  In particular, PowerLoom's classifier
> is weak in classifying based on cardinality constraints.
>
> Loom, on the other hand, being a classification-based description
> logic is quite good at classifying concepts.  As long as you are
> able to express what you wish in the more limited Loom language, it
> would be a good choice.  Also, if you want to have a system that
> supports procedural rules (production rules), then Loom would be a
> good choice.
>
> PowerLoom's strengths lie in areas outside of concept and relation
> classification.  Major benefits are a more expressive input language
> and a more expressive logical rule language (this is different from a
> procedural rule language).  PowerLoom's research release also has
> support for connections to databases, which is not available in Loom.
> (We hope to make the stable parts of that available in the next
> PowerLoom release.)
>
> PowerLoom is available in C++, Java and Common Lisp.  Loom is
> available only in Common Lisp.
>
> PowerLoom is also under active development whereas Loom is a
> finished project and will not be further developed.
>
>> Best regards,
>>
>> Kotaro Funakoshi
>> Tokyo Institute of Technology
>> _______________________________________________
>> powerloom-forum mailing list
>> [email protected]
>> http://mailman.isi.edu/mailman/listinfo/powerloom-forum
>
>