Success! Compounds merged

[Alan Ruttenberg]

Basic scheme:

(defdatarole haskey)
(equal_c compound1 key1)
(defprimconcept key1 (some haskey (string key1)))

If I use equal_c in place of defprimconcept then it takes too long to
wait. But not needed.

Test case classifies in 0.04 seconds!

Correctly finds a conflict where a compound has two keys.

I am happy. More tests tomorrow.

File attached.

-Alan

[Alan Ruttenberg]

The english makes more sense if it is written like this:

(equal_c compound1 the-things-that-are-described-by-key1)
(defprimconcept the-things-that-are-described-by-key1 (some haskey
(string key1)))

-Alan

[Ian Horrocks]

On 7 Apr 2006, at 03:15, Alan Ruttenberg wrote:

> The english makes more sense if it is written like this:
>
> (equal_c compound1 the-things-that-are-described-by-key1)
> (defprimconcept the-things-that-are-described-by-key1 (some haskey
> (string key1)))

I thought about it a bit more and I guess that you might be able to do
even better by saying:

(defprimconcept compound1 (and (some haskey1
the-things-that-are-described-by-key1) ... (some haskeyn
the-things-that-are-described-by-keyn)))
(defprimconcept the-things-that-are-described-by-key1 (and (some haskey
(string key1)) (all (inverse haskey1) compound1)))
...
(defprimconcept the-things-that-are-described-by-keyn (and (some haskey
(string keyn)) (all (inverse haskeyn) compound1)))

but given that it already takes only 0.04s, it may not be worth the
bother! In fact it may be that the improved performance is due to the
fact that the new scheme enables the reasoner to use role absorption to
automatically transform the axioms into something rather similar to the
above.

Ian