Recent changes to 28: Various Basic Calamitieshttp://sourceforge.net/p/opencyc/bugs/28/2004-03-11T12:27:46ZVarious Basic Calamities2004-03-11T12:27:46Z2004-03-11T12:27:46ZDietrich H. Fischerhttp://sourceforge.net/u/dietrichfischer/http://sourceforge.net14912b91eee2f52412ae6b69d65e574e52323569<div class="markdown_content"><p>I start with a virgin image.<br />
Create ColSub, ColSuper, and assert both to be <br />
instances of Collection.<br />
All in BaseKB.</p>
<p>A step aside:<br />
Assert: (isa ColSuper FirstOrderCollection)<br />
Assert (Monotonic-Backward): (genls ColSub ColSuper).</p>
<p>Ask: (isa ColSub FirstOrderCollection) <br />
Cannot make the system prove it.</p>
<p>!Retract: (genls ColSub ColSuper)</p>
<p>End of the step aside.</p>
<p>!Assert (Monotonic-Backward): <br />
(implies (isa ?X ColSub) (isa ?X ColSuper))</p>
<p>Ask: (genls ColSub ColSuper) <br />
Cannot make the system prove it, even if the rule <br />
(implies (isa ?X ColSub) (isa ?X ColSuper)) is given to <br />
the proof checker and allowed.</p>
<p>Ask trivially: <br />
Mt : BaseKB<br />
EL Query :<br />
(implies <br />
(isa ?X ColSub) <br />
(isa ?X ColSuper))</p>
<p>Justifications :<br />
(implies <br />
(isa ?X ColSub) <br />
(isa ?X ColSuper)) in BaseKB <br />
:ISA (isa HYP-ColSub-8884 ColSub) in Proof-<br />
HypotheticalContext-8888 </p>
<p>Note: <br />
1. (isa HYP-ColSub-8884 ColSub) is not needed in this <br />
proof.<br />
2. The parameters were: 1 transformation step, HL <br />
predicate transformation allowed.<br />
3. If this is reset to No Transformation and HL predicate <br />
transformation not allowed, <br />
then Suspended, Exhaust, No Answer<br />
4. In an image with a longer history I could not at all <br />
make the system prove <br />
this directly asserted rule. Therefore I started for this <br />
example with a virgin image.</p>
<p>Ask: <br />
(trueSentence <br />
(implies <br />
(isa ?X ColSub) <br />
(isa ?X ColSuper)))<br />
Cannot find a parameter setting such that the system <br />
finds a proof of this!</p>
<p>It has several times been complained by me <br />
that expansion axioms are not more effective than <br />
comments.</p>
<p>Therfore I add the following rule for genls (Monotonic <br />
Backward)<br />
(implies <br />
(implies (isa ?Obj ?Col1) (isa ?Obj ?Col2)) <br />
(genls ?Col1 ?Col2))</p>
<p>This is transformed by the system into the nonsense<br />
a)<br />
(implies <br />
(isa ?OBJ ?COL2) <br />
(genls ?COL1 ?COL2))<br />
b)<br />
(or <br />
(isa ?OBJ ?COL1) <br />
(genls ?COL1 ?COL2))</p>
<p>a) or b) can each be retracted without the system <br />
retracting the other.</p>
<p>Confer the typical thread in the Open Discussion Forum <br />
starting from my mail "A rule transformation bug" on <br />
2002-06-18 13:04.<br />
This implicit, but not acknowledged bug report pertained <br />
to Release 0.6 !</p>
<p>I retract a) and b) and assert: <br />
(implies<br />
(trueSentence <br />
(implies <br />
(isa ?Obj ?Col1) <br />
(isa ?Obj ?Col2)))<br />
(genls ?Col1 ?Col2))</p>
<p>This works to avoid the nonsense canonicalization.</p>
<p>Now I ask:<br />
(genls ColSub ColSuper)<br />
and cannot find a parameter setting such that I get a <br />
positive answer,<br />
even if present the pertinent rule to the proof checker <br />
on the tablet.<br />
The Query Tool answers always with No answer and <br />
suspended, Exhaust.<br />
Legacy Ask shows the pertinent rule always (!) as its <br />
last child of the inference tree,<br />
and when its antecedent is used in a "Socratic Ask" I <br />
end up as above with the Query tool.</p>
<p>Is anybody there who reads this and willing to throw <br />
some light on it?</p></div>