Representation as a Fluent
World infinitely rich: cannot model every aspect.
Fixed representations cannot cope with changing world and new
challenges.
Representation needs to evolve under machine control.
Need to change signatures – not just beliefs.
Towards a Theory of Ontology Repair
or Truthfulness Considered Harmful
Alan Bundy
University of Edinburgh
Ontology Repair System
ORS Program: repairs faulty ontologies by analysing failed multiagent plans.
Changes include abstraction and refinement of signatures,
e.g. adding arguments, changing predicates.
Physics as a Domain
Allows agents with slightly different ontologies to communicate. Historical record of ontologies, their faults, diagnosis and successors.
Plenty of examples.
Personal background in Mecho Project.
Including some formalisations.
Examples
Subtle and profound issues.
Paradox of the Bouncing Ball
Andy deSessa’s (1983)
Bouncing Ball: Where
does energy go at
moment of impact?
Essential to idealize ball as
deformable, eg spring.
Bouncing Ball Snapshots
Initially: Ball : part
Vel(Ball,0)=0, Ht(Ball,0)>0
Just before contact:
Vel(Ball,T-)>0, Ht(Ball,T-)=0
At point of contact:
Vel(Ball,T)=0, Ht(Ball,T)=0
From energy equations:
G.Ht(Ball,0) = Vel(Ball,T-)2 /2 = 0
From which ? e.g. 0=Vel(Ball,T-)>0
Representation must be fluent in reasoning.
Not just belief revision but also signature revision.
Reasoning failure can motivate and direct reasoning
refinement,
e.g. false theorem or inconsistency.
Towards formal theory.
Define most-general repair.
Avoid problems of truthfulness, semantic preservation,
inconsistency creation and undefinedness.
Properties to be Avoided
Theory
Energy of a Particle
Energy Conservation
8 o:obj, t1:time, t2:time. TE(o,t1) = TE(o,t2)
Energy of particle:
8 p:part, t:time. TE(p,t) ::= PE(p,t) + KE(p,t)
Potential Energy:
8 o:obj, t:time. PE(o,t) ::= Mass(o).G.Ht(o,t)
Kinetic Energy:
8 o:obj, t:time. KE(o,t) ::= Mass(o).Vel(o,t)2/2
Repair
Conclusion
Formal Definitions
Truthfulness: O ` ! (O) ` ()
Repair degenerates:
Repair(O,,) ::= (O` Æ ²) ) ²()
Semantics preserving: ² ! ²()
Repair degenerates:
Repair(O,,) ::= (O ` Æ ²) ! (O) ` ()
Inconsistency creating:
O ` ? Æ (O) ` ?
Repair(,) ::= O` Æ ²) ! ((O) ` () Ç ²())
In ORS is Holds(Goal,Sit)
deliberately overloaded.
=? is special case, since ² ?.
MGR(O,) ::= Repair(O,) Æ 8 ': O O. Repair(O, ') ! Th('(O)) ¾ Th((O))
Most general repairs not unique.
Abstraction Techniques
Refinement Techniques
Predicate and Function: separating them.
Vel(Ball) InVel(Ball) + AverVel(Ball)
Domain: specialise types.
Ball:obj Ball:part
Propositional: add argument to predicate.
InVel(Ball) InVel(Ball,t)
Precondition: add rule condition
[Heat(o,t) ::= Temp(o,t)] [Solid(o) ! Heat(o,t) ::= Temp(o,t)]
Predicate & Function: merging them.
Domain abstraction: Ball:spring
MStar + EStar Venus
Energy of a spring:
Domain: generalise types.
8 s:spring, t:time. TE(s,t) ::= PE(s,t) + KE(s,t) +
Ball:part Ball:string
EE(s,t)
Propositional: drop arguments.
Elastic potential energy:
AverVel(o,t) AverVel(o)
8 s:string, t:time. EE(s,t) ::=
Precondition: drop rule conditions
2
((s).(Len(s,t)-NatLen(s)) )/(2.NatLen(s))
[:Cold(o) ! Heat(o,t) ::= Temp(o,t)]
Revised energy equations:
[Heat(o,t) ::= Temp(o,t)] Permutation: change argument order
InVel(o,t) InVel(t,o)
Mass(Ball).G.Ht(Ball,0) = Mass(Ball).Vel(Ball,T-)2/2 =
Aristotle vs Galileo
Adapted from Walsh & Giunchiglia
((Ball) (Len(Ball,T)-NatLen(Ball))2)/(2.NatLen(Ball))
Aristotle conflated average and instantaneous velocity (Kuhn 1977).
Predicate Abstraction and Refinement
Propositional Abstraction & Refinement
Galileo et al discredited neo-Aristotelian physics.
Similar problems with truthfulness, inconsistency creation and undefinedness.
Propositional abstraction is truthful.
O ` Vel(Ball) = (Ht(Ball,T)-Ht(Ball,0))/T > 0
Also fix with semantic definitions.
Simple induction on proofs.
² Vel(Ball) > 0 at t=0
Predicate Abstraction:
Propositional abstraction is inconsistency creating.
Distinguish: InVel(Ball) from AverVel(Ball).
Ax(a(O)) ::= {[p] | ([p1] 2 Ax(O) Ç [p2]) 2 Ax(O) Æ O ` :[p1 Ç p2] }
P(a) 2 Ax(O) Æ : P(b) 2 Ax(O) ! (O) ` P Æ : P
Make both fluents:
Predicate Refinement:
Propositional refinement is not totally defined.
InVel(Ball,Mom), AverVel(Ball,Int).
Ax(r(O)) ::= {[p1 Ç p2] | [p] 2 Ax(O) ]}
Unconflating Heat and Temperature
P 2 Ax(O) ! P(?) 2 Ax((O))
Effect of cold:
8 o,c:obj,i:int. Adj(o,c,i) Æ Cold(c) ! Down( t. Heat(o,t),i)
Definition of Down:
Fix with Semantic Definitions
Paradox of Latent Heat
Down(f,i) ::= 8 t1,t2:mom. t1 2 i Æ t2 2 i Æ t1<t2 ! f(t1) > f(t2)
Propositional Abstraction:
Latent heat: change of heat content without change of Inference:
Ax(a(O)) ::= {[p(t)] | 9 t0. [p(t0, t)]2Ax(O) Æ O `:[9 x. p(x, t)]}
temperature.
O ` Heat(H20,Start(Freeze)) > Heat(H20,End(Freeze))
Propositional Refinement:
Black discovered in 1761.
Observation:
Ax(r(O)) ::= {[9 x. p(x, t)] | [p(t)] 2 Ax(O)}
Before Black, heat and temperature conflated.
² Heat(H20,Start(Freeze)) = Heat(H20,End(Freeze))
Blocks inconsistency:
Wiser & Carey 1983
Repair (function refinement):
: P 2 Ax(a(O)) $ :9 t0. : P(t0)2 Ax(O) Ç O` :: 9 x. P(x) $ O ` 9 x. P(x) since
Separation of conflated concepts necessary precursor
8 o:ob,t:mom. Solid(o,t) ! Heat(o,t) ::= Temp(o,t)
P(a) 2 Ax(O)
to discovery.
8 o:ob,t:mom. Liquid(o,t) ! Heat(o,t) ::= Temp(o,t)+LHF(o)
Conflation of “morning star” and “evening star” into
8 o:ob,t:mom. Gas(o,t) ! Heat(o,t) ::= Temp(o,t)+LHF(o) + LHV(o)
“Venus” in reverse direction.
cf dark matter.
© Copyright 2026 Paperzz