Professional Documents
Culture Documents
Knowledge Base
A coffee is a hot drink
With a token I can buy coffee
After buying I have a hot drink
...
¬t, c, h
b b
t, c, h b t, ¬c, h
Observations
I have got a cold coffee
I cannot buy
I bought and I got no hot drink
Observations
I have got a cold coffee
I cannot buy
I bought and I got no hot drink
¬t, c, h
b b
t, c, h b t, ¬c, h
¬t, c, h c, ¬h
b b
t, c, h b t, ¬c, h
¬t, c, h
b b
t, c, h b t, ¬c, h
¬t, c, h
t, c, h b t, ¬c, h
¬t, c, h
b b
t, c, h b t, ¬c, h
¬t, c, h
b b
t, c, h b t, ¬c, h
b
¬t, ¬c, ¬h t, ¬c, ¬h ¬t, ¬c, h
1 Preliminaries
Action Theories
3 Conclusion
Contributions
Future Work
1 Preliminaries
Action Theories
3 Conclusion
Contributions
Future Work
1 Preliminaries
Action Theories
3 Conclusion
Contributions
Future Work
1 Preliminaries
Action Theories
3 Conclusion
Contributions
Future Work
Reasoning tasks
Projection : ‘do I have a hot drink after I buy?’
Explanation : ‘I hold a coffee. I bought. Did I have a token?’
Planning : ‘how to get a hot drink?’
...
Ivan José Varzinczak (KRR–Meraka) On Action Theory Change 5 / 29
Action Theories
Knowledge bases about the dynamics of the world
Actions
Effects
Preconditions
Reasoning tasks
Projection : ‘do I have a hot drink after I buy?’
Explanation : ‘I hold a coffee. I bought. Did I have a token?’
Planning : ‘how to get a hot drink?’
...
Ivan José Varzinczak (KRR–Meraka) On Action Theory Change 5 / 29
Action Theories
Knowledge bases about the dynamics of the world
Actions
Effects
Preconditions
Reasoning tasks
Projection : ‘do I have a hot drink after I buy?’
Explanation : ‘I hold a coffee. I bought. Did I have a token?’
Planning : ‘how to get a hot drink?’
...
Ivan José Varzinczak (KRR–Meraka) On Action Theory Change 5 / 29
Action Theories in Multimodal Logic
Multimodal Logic
Propositional logic + modal operators
I [a] : every a-arrow
I hai : some a-arrow
Expressive
I Actions, state constraints, nondeterminism
Decidable
I EXPTIME-complete, though
Expressive
I Actions, state constraints, nondeterminism
Decidable
I EXPTIME-complete, though
Expressive
I Actions, state constraints, nondeterminism
Decidable
I EXPTIME-complete, though
Expressive
I Actions, state constraints, nondeterminism
Decidable
I EXPTIME-complete, though
Expressive
I Actions, state constraints, nondeterminism
Decidable
I EXPTIME-complete, though
Expressive
I Actions, state constraints, nondeterminism
Decidable
I EXPTIME-complete, though
a1
Satisfaction in a model
p, q
a1
p, ¬q q→p
p → [a1 ]¬q
a2
a1
p → ha1 i>
M : a1
p → ha2 i>
¬p, ¬q
(p ∧ ¬q) → [a2 ]⊥
a2
a1
Satisfaction in a model
p, q
a1
p, ¬q q→p
p → [a1 ]¬q
a2
a1
p → ha1 i>
M : a1
p → ha2 i>
¬p, ¬q
(p ∧ ¬q) → [a2 ]⊥
a2
a1
Satisfaction in a model
p, q
a1
p, ¬q q→p
p → [a1 ]¬q
a2
a1
p → ha1 i>
M : a1
p → ha2 i>
¬p, ¬q
(p ∧ ¬q) → [a2 ]⊥
a2
a1
Satisfaction in a model
p, q
a1
p, ¬q q→p "
p → [a1 ]¬q
a2
M : a1 a1 p → ha1 i>
p → ha2 i>
¬p, ¬q
(p ∧ ¬q) → [a2 ]⊥
a2
a1
Satisfaction in a model
p, q
a1
p, ¬q q→p "
a2
p → [a1 ]¬q "
M : a1 a1 p → ha1 i>
p → ha2 i>
¬p, ¬q
(p ∧ ¬q) → [a2 ]⊥
a2
a1
Satisfaction in a model
p, q
a1
p, ¬q q→p "
a2
"
p → [a1 ]¬q
M : a1 a1 p → ha i> "
1
p → ha2 i>
¬p, ¬q
(p ∧ ¬q) → [a2 ]⊥
a2
a1
Satisfaction in a model
p, q
a1
p, ¬q q→p "
a2
"
p → [a1 ]¬q
M : a1 a1 p → ha i> "
1
p → ha i> %
2
¬p, ¬q
(p ∧ ¬q) → [a2 ]⊥
a2
a1
Satisfaction in a model
p, q
a1
p, ¬q q→p "
a2
"
p → [a1 ]¬q
M : a1 a1 p → ha i> "
1
p → ha i> %
2
¬p, ¬q
(p ∧ ¬q) → [a2 ]⊥ "
a2
Definition
Action Theory T = S ∪ E ∪ X
Definition
Action Theory T = S ∪ E ∪ X
¬t, c, h
b b
M : t, c, h b t, ¬c, h
¬t, c, h
b b
M : t, c, h b t, ¬c, h
1 Preliminaries
Action Theories
3 Conclusion
Contributions
Future Work
¬t, c, h
b b
M : t, c, h b t, ¬c, h
¬t, c, h
b b
M : t, c, h b t, ¬c, h
¬t, c, h t, c, ¬h
b b
M : t, c, h b t, ¬c, h
¬t, c, ¬h ¬t, c, h
b b
M : t, c, h b t, ¬c, h
¬t, c, ¬h ¬t, c, h t, c, ¬h
b b
M : t, c, h b t, ¬c, h
¬t, c, h
b b
M : t, c, h b t, ¬c, h
¬t, c, h
b b
M : t, c, h b t, ¬c, h
¬t, c, h
M : t, c, h b t, ¬c, h
¬t, c, h
b b
M : t, c, h t, ¬c, h
¬t, c, h
M : t, c, h t, ¬c, h
¬t, c, h
b b
M : t, c, h b t, ¬c, h
¬t, c, h
b b
M : t, c, h b t, ¬c, h
¬t, c, h
b b
M : t, c, h b t, ¬c, h
b
¬t, ¬c, ¬h t, ¬c, ¬h ¬t, ¬c, h
¬t, c, h
b b
M : t, c, h b t, ¬c, h
¬t, c, h
b b
M : t, c, h b t, ¬c, h
b b
b
¬t, ¬c, ¬h t, ¬c, ¬h ¬t, ¬c, h
Definition
M 0 is as close to M as M 00 iff
either W −̇W 0 ⊆ W −̇W 00
or W −̇W 0 = W −̇W 00 and R −̇R 0 ⊆ R −̇R 00
Notation: M 0 M M 00
Definition
M 0 is as close to M as M 00 iff
either W −̇W 0 ⊆ W −̇W 00
or W −̇W 0 = W −̇W 00 and R −̇R 0 ⊆ R −̇R 00
Notation: M 0 M M 00
Definition
M 0 is as close to M as M 00 iff
either W −̇W 0 ⊆ W −̇W 00
or W −̇W 0 = W −̇W 00 and R −̇R 0 ⊆ R −̇R 00
Notation: M 0 M M 00
Definition
M 0 is a candidate iff
W ⊆ W0
R = R0
There is w ∈ W 0 falsifying ϕ
Definition
M 0 is a candidate iff
W ⊆ W0
R = R0
There is w ∈ W 0 falsifying ϕ
b b b b
t, c, h b t, ¬c, h M t, c, h b t, ¬c, h
¬t, ¬c, ¬h t, ¬c, ¬h ¬t, ¬c, h ¬t, ¬c, ¬h t, ¬c, ¬h ¬t, ¬c, h
b b b b
t, c, h b t, ¬c, h t, c, h b t, ¬c, h
¬t, ¬c, ¬h t, ¬c, ¬h ¬t, ¬c, h ¬t, ¬c, ¬h t, ¬c, ¬h ¬t, ¬c, h
Incomparable
Definition
M 0 is a candidate iff
W0 = W
R0 ⊆ R
There is w ∈ W 0 falsifying ϕ → hai>
Definition
M 0 is a candidate iff
W0 = W
R0 ⊆ R
There is w ∈ W 0 falsifying ϕ → hai>
¬t, c, h ¬t, c, h
t, c, h b t, ¬c, h M t, c, h t, ¬c, h
¬t, ¬c, ¬h t, ¬c, ¬h ¬t, ¬c, h ¬t, ¬c, ¬h t, ¬c, ¬h ¬t, ¬c, h
¬t, c, h ¬t, c, h
b b b
t, c, h b t, ¬c, h t, c, h t, ¬c, h
¬t, ¬c, ¬h t, ¬c, ¬h ¬t, ¬c, h ¬t, ¬c, ¬h t, ¬c, ¬h ¬t, ¬c, h
Incomparable
Definition
M 0 is a candidate iff
W0 = W
R ⊆ R0
If (w , w 0 ) ∈ R 0 \ R , then w 0 is a target (details in the JAIR paper)
There is w ∈ W 0 falsifying ϕ → [a]ψ
Definition
M 0 is a candidate iff
W0 = W
R ⊆ R0
If (w , w 0 ) ∈ R 0 \ R , then w 0 is a target (details in the JAIR paper)
There is w ∈ W 0 falsifying ϕ → [a]ψ
¬t, c, h ¬t, c, h
b b b b
t, c, h b t, ¬c, h M t, c, h b t, ¬c, h
b b
¬t, ¬c, ¬h t, ¬c, ¬h ¬t, ¬c, h ¬t, ¬c, ¬h t, ¬c, ¬h ¬t, ¬c, h
¬t, c, h ¬t, c, h
b b b b
t, c, h b t, ¬c, h t, c, h b t, ¬c, h
b
¬t, ¬c, ¬h t, ¬c, ¬h ¬t, ¬c, h ¬t, ¬c, ¬h t, ¬c, ¬h ¬t, ¬c, h
Incomparable
1 Preliminaries
Action Theories
3 Conclusion
Contributions
Future Work
Theorem
The algorithms are correct w.r.t. our semantics (details in the JAIR paper)
Theorem
Complexity is exponential, though
Nevertheless
Theorem
The algorithms always terminate
Theorem
Size of T 0 is linear in that of T
Theorem
The algorithms are correct w.r.t. our semantics (details in the JAIR paper)
Theorem
Complexity is exponential, though
Nevertheless
Theorem
The algorithms always terminate
Theorem
Size of T 0 is linear in that of T
Theorem
The algorithms are correct w.r.t. our semantics (details in the JAIR paper)
Theorem
Complexity is exponential, though
Nevertheless
Theorem
The algorithms always terminate
Theorem
Size of T 0 is linear in that of T
Theorem
The algorithms are correct w.r.t. our semantics (details in the JAIR paper)
Theorem
Complexity is exponential, though
Nevertheless
Theorem
The algorithms always terminate
Theorem
Size of T 0 is linear in that of T
1 Preliminaries
Action Theories
3 Conclusion
Contributions
Future Work
Monotonicity
T |= T 0
Preservation
If T 6|= α, then T ≡ T 0
Success
If T 6|= ⊥ and 6|= α, then T 0 6|= α
Monotonicity
T |= T 0
Preservation
If T 6|= α, then T ≡ T 0
Success
If T 6|= ⊥ and 6|= α, then T 0 6|= α
Monotonicity
T |= T 0
Preservation
If T 6|= α, then T ≡ T 0
Success
If T 6|= ⊥ and 6|= α, then T 0 6|= α
Equivalences
Contracting with equivalent formulas give the same result
Recovery
T 0 ∪ {α} |= T
Equivalences
Contracting with equivalent formulas give the same result
Recovery
T 0 ∪ {α} |= T
1 Preliminaries
Action Theories
3 Conclusion
Contributions
Future Work
Intuitive semantics
I Simple operations: add and remove
I Distance between models
I Minimal change
Intuitive semantics
I Simple operations: add and remove
I Distance between models
I Minimal change
Intuitive semantics
I Simple operations: add and remove
I Distance between models
I Minimal change
Intuitive semantics
I Simple operations: add and remove
I Distance between models
I Minimal change
1 Preliminaries
Action Theories
3 Conclusion
Contributions
Future Work
Thank you!
Thank you!