Missing consistency check because apply to structure is not always performed.
Apply to structure is only performed for symbols in the output vocabulary. Problem is that the symbolic structure might be inconsistent on atoms that are never checked as they (now) do not occur in the grounding. E.g.: !x: P(x). !x: ~P(x).
Apply to structure should at least do a consistency check for all symbols.
Comments (6)
-
-
reporter dus approximatie bestaat uit drie opties: LUP:reken symbolen al dan niet uit GWB:gebruik bounds ???: vervang symbolen (allemaal door hun bdds in de bounds) in bepaalde gevallen van combinaties van deze opties GWB+??? maar geen LUP moeten we consistentie checken. klinkt aannemelijk. maar LUP is dan niet echt meer nodig als optie of wel? mss zijn er gevallen waarin het de performantie verhoogd maar ik vermoed van niet en wat dan nog leuker zou zijn is dat zelfs outputvoc niet hoeft uitgerekend te worden tot iemand het wilt enumereren.
Probleem: apply is approximatief, terwijl de inconsistentie check exact moet zijn? Of niet?
-
reporter - changed status to resolved
Fixes
#500: if approx is true, lup decides whether apply is done in advance or only consistency checking is done and lup afterwards.→ <<cset 9472d3689ecb>>
-
reporter Fixes
#500: if approx is true, lup decides whether apply is done in advance or only consistency checking is done and lup afterwards.→ <<cset 834e36a4d469>>
-
reporter Fixes
#500: if approx is true, lup decides whether apply is done in advance or only consistency checking is done and lup afterwards.→ <<cset cd450386b175>>
-
reporter Fixes
#500: if approx is true, lup decides whether apply is done in advance or only consistency checking is done and lup afterwards.→ <<cset 87d52a0f35de>>
- Log in to comment
Disagree.
The check should only happen once you decide not to ground certain sentences because of information in the symbolic structure. As long as that information is not used, you should check nothing.