Tutorial: Nonmonotonic Logic (Day 2)

Size: px
Start display at page:

Download "Tutorial: Nonmonotonic Logic (Day 2)"

Transcription

1 Tutorial: Nonmonotonic Logic (Day 2) Christian Straßer Institute for Philosophy II, Ruhr-University Bochum Center for Logic and Philosophy of Science, Ghent University September 3, 2015 Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

2 Outline 1 Plausible Reasoning 2 Preferential / Selection Semantics (KLM, Shoham) 3 Bibliography Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

3 Topic 1 Plausible Reasoning 2 Preferential / Selection Semantics (KLM, Shoham) 3 Bibliography Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

4 Plausible vs. Defeasible Reasoning (Rescher / Vreeswijk / Prakken) deductive standard: CL Scheme: Plausible Inferences A follows nonmonotonically from B iff B A ab and there is no reason to assume ab. A follows nonmonotonically from B iff B ass A and there is a reason to assume ass. classical inference in disguise thus, e.g., contrapositable since A B ab iff B A ab; and A ass B iff B ass A explicite defeasible assumption Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

5 Plausible Reasoning can be found in, for instance,... Rescher / Manor consequence relations (Rescher and Manor (1970)) Brewka s preferred subtheories (Brewka (1989)) Makinsons Default Assumptions (Makinson (2003)) Batens Adaptive Logics and generalisations (Batens (2007); Straßer (2014)) Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

6 General Account: Premise Sets (non-defeasible) facts Σ = Σ 0, Σ d Σ d may also be stratified: defeasible premises (assumptions) Σ d = Σ 1,..., Σ n,... e.g., Σ 1 stems from the most reliable (though fallible) source, Σ 2 stems from the second most reliable (though fallible) source, etc. in Rescher/Manor consequence relations: Σ 0 = Makinsons s default assumptions: Σ 0 may be non-empty Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

7 Base Logic L with consequence relation Cn monotonic: Cn(Γ) Cn(Γ Γ ) reflexive: Γ Cn(Γ) compact: if A Cn(Γ) then A Cn(Γ ) for some finite Γ Γ cut: where Γ Cn(Γ) and A Cn(Γ Γ ), A Cn(Γ) or, Transitivity: where Γ Cn(Γ) and A Cn(Γ ), A Cn(Γ). Note: given refl. and mono., CUT iff TRANSITIVITY Suppose Γ Cn(Γ). ( ) Suppose A Cn(Γ ). By Monotonicity, A Cn(Γ Γ ). By Cut, A Cn(Γ). ( ) Suppose A Cn(Γ Γ ). By Reflexivity, Γ Γ Cn(Γ). By Transitivity, A Cn(Γ). Notation: Cn(Γ) = {A Γ A} Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

8 General Account: Consequence Relations Idea: Use (maximal) consistent subsets of Σ d. F-Inconsistent set we suppose a specific logical form F (e.g., A A) in the language L of L Γ is inconsistent if Γ B where B is of the form F otherwise Γ is consistent in the following we will write as a placeholder for formulas of the form F Maximal Consistent Subsets of Σ = Σ 0, Σ d Ξ is a maximal consistent subset of Σ iff 1 Ξ Σ d is such that Ξ Σ 0 is consistent 2 there is no Γ Σ d such that Ξ Γ and Γ Σ 0 is consistent. We write MCS(Σ) for the set of all maximal consistent subsets of Σ. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

9 Examples Let Σ = Σ 0, Σ d where Σ 0 = {s} Σ d = {s (p q), p q, r} What are the MCSs? Ξ 1 = {s (p q), r} Ξ 2 = {p q, r} Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

10 Existence of MCSs Where Ξ Σ d s.t. Σ 0 Ξ, there is a Ξ MCS( Σ 0, Σ d ) s.t. Ξ Ξ. Proof Let Σ d = {A 1,...}. Let Ξ = i 0 Ξ i where Ξ 0 = Ξ { Ξi {A Ξ i+1 = i+1 } if Ξ i {A i+1 } Σ 0 else Ξ i By induction, for each Ξ i, Ξ i Σ 0. Assume Ξ Σ 0. By compactness, there is a finite Ξ f Ξ such that Ξ f Σ 0. Thus there is a Ξ i Ξ f and by monotonicity, Ξ i Ξ 0, a contradiction. Hence, Ξ Σ 0. Where A i / Ξ, Ξ {A i } Σ 0 since Ξ i 1 {A i } Σ 0 and by monotonicity. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

11 General Account: Consequence Relations Question Do you think this also holds if we define "Γ is inconsistent iff Cn(Γ) = L"? What about... L where Cn(Γ) = Γ and L = ω? monotonic reflexive compact transitive Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

12 General Account: Consequence Relations Σ = Σ 0, Σ d Free consequences Σ free A iff Σ 0 Free(Σ) A where Free(Σ) = MCS(Σ) Universal consequences Σ A iff Σ 0 Ξ A for all Ξ MCS(Σ) Existential consequences Σ A iff Σ 0 Ξ A for some Ξ MCS(Σ) Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

13 Examples Let Σ = Σ 0, Σ d where Σ 0 = {s} Σ d = {s (p q), p q, r} What are the MCSs? Ξ 1 = {s (p q), r} Ξ 2 = {p q, r} Free consequences Free(Σ) = {r} Γ free A iff {s, r} A Notice: Syntax-Dependency Σ 0, {s p, s q, p, q, r} free p Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

14 Examples Let Σ = Σ 0, Σ d where Σ 0 = {s} Σ d = {s (p q), p q, r} What are the MCSs? Ξ 1 = {s (p q), r} Ξ 2 = {p q, r} Universal Consequences Γ A iff {s, r, p} A floating conclusion: p Question: Is there also some syntax-dependency for? Take Σ =, {p q, p} MCS(Σ ) = { {p q}, { p} } Σ q While where Σ =, {p, q, p} MCS(Σ ) = { {p, q}, { p, q} } Σ q. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

15 Examples Let Σ = Σ 0, Σ d where Σ 0 = {s} Σ d = {s (p q), p q, r} What are the MCSs? Ξ 1 = {s (p q), r} Ξ 2 = {p q, r} Existential Consequences Σ A iff A Cn(Ξ 1 {s}) Cn(Ξ 2 {s}) Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

16 General Account: Consequence Relations Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

17 What about stratified premises? Let Σ 0, Σ d where Σ d = Σ 1, Σ 2,.... Idea: Order MCSs relative to the strength of their constituting premises where strength/reliability/etc. is indirectly proportional to the index Σ i. Lexicographic Ordering (Brewka (1989), Van De Putte and Straßer (2012)) Ξ Ξ iff there is an i 1 such that 1 Ξ Σ j = Ξ Σ j for all 1 j < i, and 2 Ξ Σ i Ξ Σ i Consequence relation... relative to min (MCS(Σ)) Σ free A iff min (MCS(Σ)) A Σ A iff for all Ξ min (MCS(Σ)), Ξ A Σ A iff for some Ξ min (MCS(Σ)), Ξ A Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

18 What about stratified premises? Lexicographic Ordering (Brewka (1989), Van De Putte and Straßer (2012)) Ξ lex Ξ iff there is an i 1 such that 1 Ξ Σ j = Ξ Σ j for all 1 j < i, and 2 Ξ Σ i Ξ Σ i Example Σ 0 = {s} Σ 1 = {s p} Σ 2 = { p, q} Σ 3 = {q, r} MCSs Ξ 1 = {s p, q, r} Ξ 2 = {s p, q, r} Ξ 3 = { p, q, r} Ξ 4 = { p, q, r} order: Ξ 1 Ξ 2 Ξ 3 Ξ 4 Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

19 Caution: Other orders may not be smooth! Suppose we associate strength in direct proporitionality to the index of Σ i (e.g., premises in Σ i+1 are more reliably than premises in Σ i (i 1)). Ordering Ξ Ξ iff there there is an i 1 such that 1 Ξ Σ j = Ξ Σ j for all j > i, and 2 Ξ Σ i Ξ Σ i Example Take Σ = Σ 0, Σ 1, Σ 2,... where Σ 0 = {p i p j j > i 1} Σ i = { p i, s} for each i 1 MCSs Ξ i = {s, p i } for each i 1 Note min (MCS(Σ)) = however, we would at least expect to derive s Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

20 Some Meta-Properties Monotonicity? Clearly fails for free and universal. What about existential? important to distinguish between: Monotonicity in the defeasible premises If Σ 0, Σ d A then Σ 0, Σ d Γ A. Monotonicity in the factual premises If Σ 0, Σ d A then Σ 0 Γ, Σ d A. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

21 Weakening of Monotonicity? What about Cautious Monotonicity in the free and universal case? Recall: Cautious Monotonicity If Σ A and Σ B then Σ {A} B. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

22 Cumulativity If Σ A : Σ B iff Σ {A} B This is: Cautious monotonicity and Cut In our case two options: in defeasible premises: If Σ 0, Σ d A : Σ 0, Σ d B iff Σ 0, Σ d {A} B in factual premises: If Σ 0, Σ d A : Σ 0, Σ d B iff Σ 0 {A}, Σ d B Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

23 Where Σ = Σ 0, Σ d, Σ A, π : MCS( Σ 0, Σ d ) MCS( Σ 0, Σ d {A} ), Λ Λ {A} is a bijection. Sketch of the Proof In case A Σ d the proposition is trivial. Suppose thus that A / Σ d. ( ) Let Ξ MCS(Σ). We show that Ξ {A} MCS( Σ 0, Σ d {A} ). Note that Ξ Σ 0 A and thus Σ 0 Ξ {A} by cut and ( ). Assume that there is a Ξ Ξ {A} such that Ξ MCS(Σ 0, Σ d {A}). Thus, Σ d Ξ \ {A} Ξ. Since Σ 0 (Ξ \ {A}), this is a contradiction to ( ). ( ) Let Ξ MCS( Σ 0, Σ d {A} ). Clearly, Σ 0 (Ξ \ {A}). Assume that there is a Ξ MCS( Σ 0, Σ d ) such that Ξ \ {A} Ξ. By the supposition, Σ 0 Ξ A and hence Σ 0 Ξ {A}. This is a contradiction to ( ) since Ξ Ξ {A}. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

24 The Cumulativity of : defeasible premises Recall: Where Σ A, π : MCS( Σ 0, Σ d ) MCS( Σ 0, Σ d {A} ), Λ Λ {A} is a bijection. Where Σ 0, Σ d A: Σ 0, Σ d B iff Σ 0, Σ d {A} B. Proof In case A Σ d the proposition is trivial. Suppose thus that A / Σ d. Suppose Σ 0, Σ d A. Suppose Σ 0, Σ d B. Let Ξ MCS( Σ 0, Σ d {A} ). Hence, Ξ \ {A} MCS( Σ 0, Σ d ) and thus Σ 0 (Ξ \ {A}) B. By Mono., Σ 0 Ξ B. Suppose Σ 0, Σ d {A} B. Let Ξ MCS( Σ 0, Σ d ). Hence, Ξ {A} MCS( Σ 0, Σ d ) and Σ 0 Ξ A. Also, Σ 0 Ξ {A} B. By cut, Σ 0 Ξ B. Thus, Σ 0, Σ d B. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

25 Stronger form of cumulativity? Do you think the antecedent of the previous result can be weakened? We d need to weaken the antecedent of our lemma: Where Σ A, π : MCS( Σ 0, Σ d ) MCS( Σ 0, Σ d {A} ), Λ Λ {A} is a bijection. Suggestions? Here s one way Suppose we have a supraclassical core logic (with cl.neg. and = p p), then Where Σ A, π : MCS( Σ 0, Σ d ) MCS( Σ 0, Σ d {A} ), Λ Λ {A} is a bijection. proof remains nearly the same (check this!) Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

26 Here it is... (changes in red) Where L is supraclassical and Σ A, π : MCS( Σ 0, Σ d ) MCS( Σ 0, Σ d {A} ), Λ Λ {A} is a bijection. Sketch of the Proof In case A Σ d the proposition is trivial. Suppose thus that A / Σ d. ( ) Let Ξ MCS(Σ). We show that Ξ {A} MCS( Σ 0, Σ d {A} ). Note that Ξ Σ 0 A and thus Σ 0 Ξ {A}. Assume that there is a Ξ Ξ {A} such that Ξ MCS(Σ 0, Σ d {A}). Thus, Σ d Ξ \ {A} Ξ. Since Σ 0 (Ξ \ {A}), this is a contradiction to ( ). ( ) Let Ξ MCS( Σ 0, Σ d {A} ). Clearly, Σ 0 (Ξ \ {A}). Assume that there is a Ξ MCS( Σ 0, Σ d ) such that Ξ \ {A} Ξ. By the supposition, Σ 0 Ξ A and hence Σ 0 Ξ {A}. This is a contradiction to ( ) since Ξ Ξ {A}. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

27 The Cumulativity of free : defeasible premises Where Σ 0, Σ d free A: Σ 0, Σ d free B iff Σ 0, Σ d {A} free B. Proof In case A Σ d the proposition is trivial. Suppose thus that A / Σ d. Note that by our lemma we immediately get: ( ) MCS( Σ 0, Σ d {A} ) = MCS(Σ) {A}. Suppose Σ 0, Σ d free B and hence MCS( Σ0, Σ d ) B. By ( ) and monotonicity, MCS( Σ0, Σ d {A} ) B. Hence, Σ 0, Σ d {A} free B. Suppose Σ 0, Σ d {A} free B and hence MCS( Σ 0, Σ d {A} ) B. By ( ), MCS( Σ0, Σ d ) {A} B. Since MCS(Σ) A, by cut MCS( Σ0, Σ d ) B. Thus Σ 0, Σ d free B. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

28 Strengthenings Strengthening 1 Where Σ 0, Σ d A: Σ 0, Σ d free B iff Σ 0, Σ d {A} free B. Strengthening 2, where L is supraclassical Where Σ 0, Σ d A: Σ 0, Σ d free B iff Σ 0, Σ d {A} free B. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

29 If Σ 0, Σ d A: Σ 0, Σ d B iff Σ 0 {A}, Σ d B ( {, free }). This follows immediately from the following Lemma If Σ 0, Σ d A then MCS( Σ 0, Σ d ) = MCS( Σ 0 {A}, Σ d ). Proof: Suppose Σ 0, Σ d A. ( ) Let Ξ MCS( Σ 0, Σ d ). Thus, Ξ Σ 0 A and hence Σ 0 Ξ {A} by ( ) and cut. Assume there is a Ξ MCS(Σ 0 {A}, Σ d ) such that Ξ Ξ. Hence, Σ 0 {A} Ξ and hence Σ 0 Ξ by mono.. Since also Ξ Σ d this contradicts ( ). Let ( ) Ξ MCS( Σ 0 {A}, Σ d ). Assume there is a Ξ MCS( Σ 0, Σ d ) for which Ξ Ξ. Hence, Σ 0 Ξ A and Σ 0 Ξ. Thus also Σ 0 {A} Ξ in contradiction to ( ). Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

30 Fixed Point Let Cn(Σ) = df {A Σ A}. Fixed Point Property Cn(Σ) = Cn(Cn(Σ)). Now: two versions (where {, free }) 1 relative to defeasible premises: Cn( Σ 0, Σ d ) = Cn( Σ 0, Σ d Cn( Σ 0, Σ d ) ) 2 relative to factual premises: Cn( Σ 0, Σ d ) = Cn( Σ 0 Cn( Σ 0, Σ d ), Σ d ) Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

31 Fixed Point follows from Cumulativity: ( ) All we need is (where {, free }): 1 Where Γ Cn ( Σ 0, Σ d ): Cn ( Σ 0, Σ d ) = Cn ( Σ 0, Σ d Γ ) 2 Where Γ Cn ( Σ 0, Σ d ): Cn ( Σ 0, Σ d ) = Cn ( Σ 0 Γ, Σ d ) Corollary: If satisfies ( ), Cn has the fixed point property. Proof: Follows immediately, just let Γ = Cn (Σ). Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

32 Proof of ( ) Our previous lemmas can easily be strengthened to: 1 Where Γ Cn ( Σ 0, Σ d ): π : MCS( Σ 0, Σ d ) MCS( Σ 0, Σ d Γ ), Λ Λ Γ is a bijection. 2 Where Γ Cn ( Σ 0, Σ d ): MCS( Σ 0, Σ d ) = MCS( Σ 0 Γ, Σ d ). Fixed Point follows from Cumulativity: The proofs for the following strong forms of cumulativity are easily adjusted (where {, free }): 1 Where Γ Cn : Cn ( Σ 0, Σ d ) = Cn ( Σ 0, Σ d Γ ) 2 Where Γ Cn : Cn ( Σ 0, Σ d ) = Cn ( Σ 0 Γ, Σ d ) Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

33 The Resolution Theorem: in the defeasible premises Do we have (supposing we have the Res.Thm. for ) Σ 0, Σ d A B implies Σ 0, Σ d {A} B? Σ 0, Σ d free A B implies Σ 0, Σ d {A} free B? for : NOPE Consider Σ =, { p}, A = p and B = q. Then Σ p q while, { p, p} q. for free : NOPE same counterexample Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

34 The Resolution Theorem for facts? Do we have Σ 0, Σ d A B implies Σ 0 {A}, Σ d B? Σ 0, Σ d free A B implies Σ 0 {A}, Σ d free B? Nope: Take Σ =, { p} and Σ + = {p}, { p}. Then Σ p q while Σ + q. And Σ free p q while Σ + free q. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

35 The Deduction Theorem for defeasible premises Question: Σ 0, Σ d {A} B implies Σ 0, Σ d A B? Σ 0, Σ d {A} free B implies Σ 0, Σ d free A B? We suppose in the following discussion 1 we have the deduction theorem on the level of L: Γ {A} B implies Γ A B. 2 explosion: A for all A L. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

36 The Deduction Theorem for defeasible premises, the universal case Σ 0, Σ d {A} B implies Σ 0, Σ d A B? Proof Suppose Σ 0, Σ d {A} B and let Ξ MCS(Σ 0, Σ d ). We have two case: (1) Ξ {A} MCS( Σ 0, Σ d {A} ) or (2) Ξ MCS(Σ 0, Σ d {A}) and Ξ Σ 0 {A} and thus Ξ Σ 0 {A} B (by explosion and trans). 1 Then by the supposition Σ 0 Ξ {A} B and hence Σ 0 Ξ A B by the deduction theorem. 2 Σ 0 Ξ A B by the deduction theorem. Thus, Ξ A B and Σ 0, Σ d A B. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

37 The Deduction Theorem for defeasible premises, the free case Σ 0, Σ d {A} free B implies Σ 0, Σ d free A B Proof Suppose Σ 0, Σ d {A} free B and hence Σ 0 MCS( Σ 0, Σ d {A} ) B. It is easy to see that MCS( Σ0, Σ d {A} ) MCS( Σ 0, Σ d ) {A}. By monotonicity, Σ 0 MCS( Σ 0, Σ d ) {A} B and by the deduction theorem, Σ 0 MCS( Σ 0, Σ d ) A B. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

38 The Deduction Theorem for Facts, the universal case Σ 0 {A}, Σ d B implies Σ 0, Σ d A B First a useful Lemma If Ξ MCS( Σ 0, Σ d ) and Σ 0 Ξ {A} then Ξ MCS( Σ 0 {A}, Σ d ). (Trivial) Proof Since Σ 0 Ξ {A} there is a Ξ Ξ such that Ξ MCS( Σ 0 {A}, Σ d ). Thus, Ξ Σ 0 and by the maximality of Ξ, Ξ = Ξ. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

39 The Deduction Theorem for Facts, the universal case Σ 0 {A}, Σ d B implies Σ 0, Σ d A B Our useful Lemma If Ξ MCS( Σ 0, Σ d ) and Σ 0 Ξ {A} then Ξ MCS( Σ 0 {A}, Σ d ). Proof of the deduction theorem Suppose Σ 0 {A}, Σ d B. Let Ξ MCS(Σ 0, Σ d ). If Σ 0 Ξ {A} also Σ 0 Ξ {A} B by explosion and transitivity. By the deduction theorem, Σ 0 Ξ A B. If Σ 0 Ξ {A} then by the Lemma, Ξ MCS( Σ 0 {A}, Σ d ) and thus Σ 0 Ξ B. By monotonicity, Σ 0 Ξ {A} B. By the deduction theorem, Σ 0 Ξ A B. ChristianAltogether, Straßer (RUB, UGENT) Σ, Σ Tutorial: A Nonmonotonic B. Logic (Day 2) September 3, / 96

40 The Deduction Theorem for facts, the free case Do we have Σ 0 {A}, Σ d free B implies Σ 0, Σ d free A B? Nope: here s a counter-example Let Σ 0 = {p q, r q} Σ d = { p, q} We have Σ 0, Σ d free p r while Σ 0 {p}, Σ d free r To see this notice that MCS( Σ 0, Σ d ) = {{ p}, { q}} MCS( Σ 0 {p}, Σ d ) = {{ q}} Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

41 Rational Monotonicity for defeasible premises: the universal case If Σ 0, Σ d B and Σ 0, Σ d A then Σ 0, Σ d {A} B? Nope! Take: Σ 0 = Σ d = {r, p q r, (p r) q, p q} A = p and B = q Σ 0, Σ d q Σ 0, Σ d p but Σ 0, Σ d {p} q. Do you see why? MCS( Σ 0, Σ d ) = 1 {(p r) q, r, p q} 2 {(p r) q, p q r} MCS( Σ 0, Σ d {p} ) = 1 {(p r) q, r, p q} 2 {(p r) q, p q r, p} 3 {r, p, (p r) q} Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

42 Rational Monotonicity for factual premises: the universal case If Σ 0, Σ d B and Σ 0, Σ d A then Σ 0 {A}, Σ d B? Nope! Take: Σ 0 = Σ d = {r, p q r, (p r) q, p q} A = p and B = q Σ 0, Σ d q Σ 0, Σ d p but Σ 0 {p}, Σ d q. Do you see why? MCS( Σ 0, Σ d ) = 1 {(p r) q, r, p q} 2 {(p r) q, p q r} MCS( Σ 0 {p}, Σ d ) = 1 {r, (p r) q} Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

43 Rational Monotonicity relative to defeasible premises: the free case If Σ 0, Σ d free B and Σ 0, Σ d free A then Σ 0, Σ d {A} free B? Nope! Take: Σ 0 = Σ d = {r, p q r, (p r) q, p q} A = p q and B = (p r) q Σ 0, Σ d free (p r) q Σ 0, Σ d free (p q) but Σ 0, Σ d {p q} free (p r) q. Do you see why? MCS( Σ 0, Σ d ) = 1 {(p r) q, r, p q} 2 {(p r) q, p q r} MCS( Σ 0, Σ d {p q} ) = 1 {(p r) q, r, p q} 2 {(p r) q, p q r, p q} 3 {r, p q} Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

44 Rational Monotonicity relative to strict premises: the free case If Σ 0, Σ d free B and Σ 0, Σ d free A then Σ 0 {A}, Σ d free B? Nope! Take: Σ 0 = Σ d = {r, p q r, (p r) q, p q} A = p q and B = (p r) q Σ 0, Σ d free (p r) q Σ 0, Σ d free (p q) but Σ 0 {p q}, Σ d free (p r) q. Do you see why? MCS( Σ 0, Σ d ) = For defeasible premises 1 {(p r) q, r, p q} 2 {(p r) q, p q r} MCS( Σ 0 {p q}, Σ d ) = 1 {r} 2 {p q r, (p r) q} Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

45 Other forms of Rationality Negation Rationality Γ {A} B Γ { A} B Γ B or positively: If Γ B then Γ {A} B or Γ { A} B. Disjunctive Rationality Γ {A} C Γ {B} C Γ {A B} C or positively If Γ {A B} C then Γ {A} C or Γ {B} C. We have again two versions: relative to factual and to defeasible premises. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

46 Disjunctive Rationality implies Negation Rationality Replacement of equivalents Let be a consequence relation between (L) (L) and L. satisfies replacement of equivalents iff, where A A, 1 Σ 0 {A}, Σ d B iff Σ 0 {A }, Σ d B 2 Σ 0, Σ d {A} B iff Σ 0, Σ d {A } B. Do we have it? Where {,, free }, satisfies replacement of equivalents. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

47 Disjunctive Rationality implies Negation Rationality Where satisfies replacement of equivalents, Disjunctive Rationality in the factual [defeasible] premises implies negation rationality in the factual [defeasible] premises. Proof 1 Suppose Σ 0 {A}, Σ d C and Σ 0 {B}, Σ d C implies Σ 0 {A B}, Σ d C, for all A, B and C. 2 Suppose also Σ 0 {A B}, Σ d C and Σ 0 {A B} C. 3 By 1, Σ 0 {(A B) (A B)}, Σ d C. 4 Since A ((A B) (A B)), Σ 0 {A}, Σ d C. Proof for defeasible premises is analogous. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

48 Counter-example for Negation Rationality for Let Σ = Σ 0, Σ d be such that Σ 0 = and Σ d = { r s, r ( q t), s r, s (q t)}. Note that MCS( Σ 0, Σ d {p} ) 1 { r s, r ( q t), s (q t), p} 2 { s r, r ( q t), s (q t), p} Hence, Σ 0 {p} t. MCS( Σ 0, Σ d {p q} ) 1 { s r, r ( q t), s (q t)} 2 { r s, r ( q t), s (q t)} 3 { r s, s (q t), p q} MCS( Σ 0 {p}, Σ d ) 1 { r s, r ( q t), s (q t)} 2 { s r, r ( q t), s (q t)} Hence, Σ 0 {p} t. MCS( Σ 0 {p q}, Σ d ) 1 { s r, r ( q t), s (q t)} 2 { r s, s (q t)} Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

49 Recapture Let Σ = Σ 0, Σ d. Where Σ 0 Σ d, Σ A iff Σ 0 Σ d A ( {, free, }) Reason: there is only one MCS, namely Σ d. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

50 An overview free Ded.Thm.(s) Ded.Thm.(d) Res.Thm(s) Res.Thm(d) CM(s) CM(d) Cut(s) Cut(d) fixed-point(s) fixed-point(d) RM(s) RM(d) Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

51 From MCSs to Semantic Selections We now suppose that L has an adequate semantics with no inconsistent models. Σ A iff Σ A (where Σ A iff for all M M(Σ), M = A) M M({ }) = for all of the form F Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

52 From MCSs to Semantic Selections Given Σ = Σ 0, Σ d, where M M(Σ 0 ), d(m) = {A Σ d M = A} M m (Σ) = {M M(Σ 0 ) there are no M M(Σ 0 ) for which d(m) d(m )} Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

53 Lemma 1 Σ 0 Ξ is L-consistent iff M(Σ 0 Ξ) 2 MCS(Σ 0, Σ d ) = {d(m) M M m ( Σ 0, Σ d )} Proof of 1 Σ 0 Ξ is inconsistent iff Σ 0 Ξ iff Σ 0 Ξ iff M(Σ 0 Ξ) = Proof of 2 Ad 2: ( ) Suppose Ξ MCS(Σ 0, Σ d ). Assume there is no M M(Σ 0 ) for which d(m) Ξ. Hence, M(Σ 0 Ξ) = and thus Σ 0 Ξ. Since then Σ 0 Ξ : contradiction to ( ). Assume there is a M M(Σ 0 ) for which d(m) Ξ. Then, M(Σ 0 d(m)) and hence Σ 0 d(m). Since then Σ 0 d(m) : contradiction to ( ). Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

54 From MCSs to Semantic Selections Definition, where Σ = Σ 0, Σ d Σ m A iff for all M M m (Σ), M = A Universal Consequence: Semantic Selections, where Σ = Σ 0, Σ d Proof Σ A, iff Σ A iff Σ m A for all Ξ MCS(Σ), Σ 0 Ξ A, iff for all Ξ MCS(Σ), Σ 0 Ξ A, iff for all Ξ MCS(Σ) and for all M M(Σ 0 Ξ), M = A, iff for all M M m (Σ), M = A, iff Σ m A. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

55 Definition, where Σ = Σ 0, Σ d Σ f A iff for all M M f (Σ), M = A where M f (Σ) = {M M(Σ 0 ) A d(m) iff there is a M M m (Σ) such that A d(m )}. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

56 Free Consequence: Semantic Selections, where Σ = Σ 0, Σ d Proof Σ free A iff Σ f A Σ free A, iff Σ 0 Free(Σ 0, Σ d ) A, iff Σ 0 MCS(Σ) A, iff Σ 0 MCS(Σ) A, iff for all M M(Σ 0 MCS(Σ)), M = A, iff for all M M(Σ 0 ) such that d(m) MCS(Σ), M = A, iff for all M M(Σ 0 ) such that d(m) {d(m ) M M m (Σ)}, M = A, iff Σ f A. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

57 From MCSs to Semantic Selections Definition: where Σ = Σ 0, Σ d Σ e A iff there is a M M m (Σ) such that for all M M m (Σ) for which d(m ) = d(m), M = A. Existential Consequence Σ A iff Σ e A Proof Σ A, iff, there is a Ξ MCS(Σ) such that Σ 0 Ξ A, iff, there is a M M m (Σ) such that Σ 0 d(m) A, iff, there is a M M m (Σ) such that for all M M m (Σ), M = A (note that there are no M M(Σ 0 ) for which d(m ) d(m)!) Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

58 Adaptive Logics The basic idea: semantics 1 Take a core logic L (the lower limit logic ) that is a Tarski Logic (its consequence relation is reflexive, transitive and monotonic) supraclassical (it has a classical and a classical ) 2 declare some logic form as being abnormal : formulas of that form are abnormalities let Ω be the set of all formulas that have the abnormal form 3 define a consequence relation by means of selecting L-models that are sufficiently normal i.e., that validate not too many abnormalities Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

59 Inconsistency-adaptive logics 1. Lower limit logic: paraconsistent logic CLuN positive fragment of classical logic and A A incl. classical negation (A 1) A (B A) (A 2) (A (B C)) ((A B) (A C)) (A 3) ((A B) A) A (A 1) (A B) A (A 2) (A B) B (A 3) A (B (A B)) (A 1) A (A B) (A 2) B (A B) (A 3) (A C) ((B C) ((A B) C)) (A 1) (A B) (A B) (A 2) (A B) (B A) (A 3) (A B) ((B A) (A B)) (A 1) (A A) A (A 2) A ( A B) (A 3) A A (A 3) A A If we add de Morgan and double-negation intro/elim then we get CLuNs. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

60 Inconsistency-adaptive logics 2. Abnormalities -contradictions: A A Ω = {A A A L} 3. Interpret premises as normal as possible i.e., interpret (A A) as true as much as possible this allows for defeasible disjunctive syllogism: If A B and A and we assume (A A), then B. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

61 Example: Let Γ = {p q, r s, p, r, r}. We cannot assume (r r) since r r is derivable. So, s cannot be derived via disjunctive syllogism. However, we can safely assume (p p). Thus, we can defeasibly apply disjunctive syllogism and derive q. Semantically, we choose models which minimise contradictions : models that validate p p are sorted out for all remaining models M: M = (p p) M = p M = p M = p q M = q Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

62 More complicated example Let Γ = {p s, r s, p, r, r p}. Now we can derive the minimal disjunction of contradictions: (r r) (p p). Should s be derivable? remember: floating conclusion different reasoning strategies Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

63 Inconsistency-adaptive logics Option 1: minimal abnormality strategy Where Γ L, where M is a L-model, Ω(M) = {A Ω M = A} M m.a. L (Γ) = {M M L (Γ) there is no M M L (Γ) for which Ω(M ) Ω(M)} Γ m.a. A iff for all M M m.a. L (Γ), M = A. Note Γ m.a. A iff Γ, Ω A Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

64 Example: Minimal abnormality Γ = {p s, r s, p, r, r p} model p p r r s s M M M M M M Thus, Γ m.a. s. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

65 Inconsistency-adaptive logics Option 2: reliability strategy Where Γ L, M rel L (Γ) = {M M L(Γ) A Ω(M) iff there is a M M m.a. L (Γ) such that A Ω(M )} Γ rel L A iff for all M Mrel, M = A. L Note Γ rel L A iff Γ, Ω free A Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

66 Example: Reliability Γ = {p s, r s, p, r, r p} model p p r r s s M M M M M M Thus, Γ rel. s. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

67 Dynamic Proof Theories for Adaptive Logics line number justification l A R; l, l We have 3 generic rules: formula assumption / condition Premise introduction (PREM) Where A Γ, introduce A on the empty condition. Unconditional Rule (RU) Where A 1,..., A n L B, l 1 A 1 J l n A n J n n l B RU;l 1,..., l n 1... n Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

68 Dynamic Proof Theories for Adaptive Logics Conditional Rule (RC) Where A 1,..., A n L B Θ and Θ Ω, l 1 A 1 J l n A n J n n l B RC;l 1,..., l n 1... n Θ Example: Γ = {p s, p, p}. 1 p s PREM 2 p PREM 3 s 1,2;RC {p p} 4 p PREM 5 p p 2,4;RU This calls for retraction. Differs with the strategy! Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

69 Dynamic Proof Theories for Adaptive Logics, Retraction The unreliable abnormalities Let U s = { Ω is derived on the condition at some line l and there is no such that is derived on the condition at stage s members of U s are the unreliable abnormalities at stage s Marking: reliability strategy Line l with condition is marked at stage s of a proof, iff U s. Example 1 p s PREM 2 p PREM 3 s 1,2;RC {p p} 4 p PREM 5 p p 2,4;RU Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

70 More complicated example Γ = {p s, r s, p, r, r p}. 1 p s PREM 2 p PREM 3 s 1,2;RC {p p} 4 r p PREM 5 r PREM 6 (p p) (r r) 2,4,5;RU 7 r s PREM 8 s 5,7;RC {r r} U 8 = {p p, r r} Minimal abnormality: multiple arguments matter either p p holds and r r not then the assumption of line 8 is OK or, vice versa, r r holds and p p not then the assumption of line 3 is OK Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

71 Final derivability A is finally derived at a line l at stage s of a proof from Γ iff 1 l is unmarked at stage s 2 for all extensions of the proof in which l is marked there is a further extension such that l is unmarked. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

72 Flip/Flop Problem: CLuNs 1 q PREM 2 p p PREM 3 q ( q q) 1;RU 4 q 3;RU 5 q r PREM 6 r 4,5;RC {q q} 7 p r 2,6;RU {q q} 8 p r 2;RU 9 (p r) 8;RU 10 (p r) (p r) 7,9;RU {q q} 11 [(p r) (p r)] [q q] 10;RA Solution: restrict abnormalities to contradictions in atoms. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

73 Topic 1 Plausible Reasoning 2 Preferential / Selection Semantics (KLM, Shoham) 3 Bibliography Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

74 Preferential / Selection Semantics We have seen already an example of these with Adaptive Logics. they have been systematically investigated by Shoham (Shoham (1987)) by Kraus, Lehmann and Magidor (Kraus et al. (1990)) Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

75 The System P Various characterisations can be found, here s one of them: A B A C Left Logical Equ. B C Right Weakening A B C A C B Two perspectives Reflexivity A A 1 as a consequence relation And A B A C 2 as operator in the object A B C language ( conditional logics of Or A C B C normality ) A B C CM A B A C A B C Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

76 Some derived properties (S) [Ded.Thm.]: A B C implies A B C 1 Suppose A B C. 2 By Right Weakening, A B B C. 3 A (A B) (A B) 4 A B A B by Reflexivity and thus A B B C by Right Weakening 5 A B C by OR (applied to 2 and 4) and LLE (in view of 3). (Cut): A B and A B C implies A C. 1 Suppose A B and A B C. 2 Thus, A B C by S. 3 A B (B C) by AND (applied to 1 and 2) 4 A C by Right Weakening. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

77 The (Standard) Semantics of P: Preferential models W,, v where W is a set of points (worlds) v : W (Atoms) is an assignment is a strict partial order such that for each A, {w w = A} is smooth NOPE! Consequence relation: Preferential Closure let min (A) = {w W w = A and for all w W such that w = A, w w} M = A B iff for all w min (A), w = B A P B iff for all preferential models M, M = A B. Alternative: A P B iff A B is derivable from system P. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

78 Central characterisational result: is a preferential consequence relation iff there is a preferential model M that characterises it (ie., A B iff M = A B.) Task Build a preferential model W,, v that characterises a consequence relation that contains all of {p b, b f, p f } with the four worlds: world p f w w w w Since W and v are already determined, all you need to do is to specify, ie., to order the worlds in a good way. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

79 The System R, Rational consequence relations and models R is P plus Rational Monotonicity A C A B A B C Semantics W,, v where W,, v is a preferential model is a modular order (i.e., we have a ranking function) Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

80 A Problem with Irrelevance Model M {p b, b f, p f } R p a f. M = p b M = p f M = b f but M = p a f. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

81 A Problem with Irrelevance In fact, despite RM, where K is a conditional knowledge base, K P A B iff K R A B. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

82 Enters: Rational Closure Interpret worlds as normal as possible relative to and K. i.e., drop worlds as low as possible min (p a) min (p) thus: M = p a f Idea: order rational consequence relations according to how normal they interpret worlds and pick out the minimal one. Formally, where A < B iff A B B 1 2 iff 1 there is a (A, B) 2 \ 1 s.t. for all C for which C < 1 A and for all D s.t. C 1 D, also C 2 D. 2 for all C, D: if (C, D) 1 \ 2, there is a (A, B) 2 \ 1 s.t. A < 2 C. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

83 Enters: Rational Closure Formally: we write A < B iff A B B 1 2 iff 1 there is a (A, B) 2 \ 1 s.t. for all C for which C < 1 A and for all D s.t. C 1 D, also C 2 D. 2 for all C, D: if (C, D) 1 \ 2, there is a (A, B) 2 1 s.t. A < 2 C. 1 vs. 2 vs : take (p a, f ) 2 \ : take (p, a) 3 \ 3 Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

84 Existence Lemma: Invariance under renaming Let f : Atoms Atoms be a bijection. 1 where 1 and 2 are rational consequence relations, 1 2 implies f ( 1 ) f ( 2 ). 2 f (RC(K)) = RC(f (K)) 3 if K = f (K) then RC(K) = f (RC(K)) (if RC(K) exists) Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

85 RC does not always exist Take K = {p n p n+2, p n+2 p n n N N + 0 }. then p n+2 < p n for all n for all odd n:... p n+4 < p n+2 < p n <... for all even n:... p n+4 < p n+2 < p n <... Suppose we have the rational closure of K Four cases: 1 p k < p n for all odd k and all even n 2 p k < p n for all even k and all odd n 3 there are two even n, m and an odd k such that p n < p k < p m 4 there are two odd n, m and an even k such that p n < p k < p m However, Ad 1/2: let f (m) = m + 1. Then f (K) = K, but f ( ). { l l is odd Ad 3/4: let f (l) = Now p l m + n l is even m < p k in f ( ). Contradiction. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

86 Drowning Problem K Note that: 1 K P penguin haswings 2 K R penguin haswings 3 K RC penguin haswings Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

87 Drowning Problem Here are our possible worlds: w 1 w 2 w 3 w 4 w 5 w 6 w 7 w 8 w 9 w 10 w 11 w 12 w 13 w 14 w 15 w p b w f The model of the rational closure Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

88 Swedes swedes blond swedes tall Should we conclude for the short swede Peter that he is blond? this is blocked in RC note that RM cannot be used to conclude this since: swedes short (where: short implies tall) Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

89 Various Semantic Characterisations From the abstract point of view S = W, v, D,, π where π : (W ) D S = A B iff π([a B]) preferable to π([a B]) in view of. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

90 π S = A B iff... structures possibility measure π([a]) = 0 or possibilistic π : (W ) [0, 1] π([a B]) > π([a B]) ordinal ranking fkt. κ([a]) = or ordinal ranking κ : (W ) {0, 1,..., } κ([a B]) < κ([a B]) plausibility meas. Pl([A]) = or plausibility Pl : (W ) D Pl([A B]) > Pl([A B]) possibility measures: Dubois and Prade (1990) 0: impossible states, 1: necessary states ordinal ranking functions Goldszmidt and Pearl (1992) κ([a]): level of surprise if A were to hold maximal surprise plausibility measures Friedman and Halpern (1996) D partially ordered domain with and some simple constraints, e.g., Pl(X ) = Pl(Y ) implies Pl(X Y ) = result in qualitative plausibility structures. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

91 Various Semantic Characterisations Where K is a conditional knowledge base: 1 K P A B iff 2 S = A B for all preferential structures S which validate K iff 3 S = A B for all possibilistic structures S which validate K iff 4 S = A B for all ordinal ranking structures S which validate K iff 5 S = A B for all qualitative plausibility structures S which validate K Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

92 Probabilities, Naive idea: A B holds in a structure if P(B A) > τ where τ is a threshold value (e.g., 1 2 or 2 3, etc.). We loose properties! What about A C (cut)?! A B (A = being a Pennsylvanian Dutch, B = being a native speaker of German) A B C (C = being born in Germany) Take the distribution P(B A) = P(A B) P(A) = = 3 4 P(C A B) = P(A B C) P(A B) = = 2 3 P(C A) = P(A C) P(A) = = 1 2 Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

93 Better idea: go for the limes ɛ-semantics, idea: (Adams (1975); Pearl (1989)) K A A B holds if P(B A) is converges to 1 Formally, for each ɛ ]0, 1] there is a δ ]0, 1] such that P(A B) 1 ɛ in all probability assignments that satisfy P(C D) 1 δ and P(C) > 0 for all C D K This results in yet another adequate semantics for P. Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

94 Other probabilistic characterisations Big stepped probabilities A B iff P(B A) > 1 2 wait, but how? order W linearly by and request P({w}) > {P({w }) w w} De Finetti (coherence-based probabilities) take conditional probabilities as primitive Gilio (2002) similar: Popper-functions Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

95 Topic 1 Plausible Reasoning 2 Preferential / Selection Semantics (KLM, Shoham) 3 Bibliography Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

96 Bibliography Adams, E. W.: 1975, The Logic of Conditionals. D. Reidel Publishing Co. Batens, D.: 2007, A Universal Logic Approach to Adaptive Logics. Logica Universalis 1, Brewka, G.: 1989, Preferred Subtheories: An Extended Logical Framework for Default Reasoning.. In: IJCAI, Vol. 89. pp Dubois, D. and H. Prade: 1990, An introduction to possibilistic and fuzzy logics. In: Readings in Uncertain Reasoning. pp Friedman, N. and J. Y. Halpern: 1996, Plausibility measures and default reasoning. Journal of the ACM 48, Gilio, A.: 2002, Probabilistic reasoning under coherence in System P. Annals of Mathematics and Artificial Intelligence 34(1-3), Goldszmidt, M. and J. Pearl: 1992, Rank-based Systems: A Simple Approach to Belief Revision, Belief Update, and Reasoning about Evidence and Actions. In: Proceedings of the Third International Conference on Knowledge Representation and Reasoning. pp Kraus, S., D. Lehman, and M. Magidor: 1990, Nonmonotonic Reasoning, Preferential Models and Cumulative Logics. Artifical Intelligence 44, Makinson, D.: 2003, Bridges between classical and nonmonotonic logic. Logic Journal of IGPL 11(1), Pearl, J.: 1989, Probabilistic semantics for nonmonotonic reasoning: a survey. In: Proceedings of the first international conference on Principles of knowledge representation and reasoning. San Francisco, CA, USA, pp Rescher, N. and R. Manor: 1970, On inference from inconsistent premises. Theory and Decision 1, Shoham, Y.: 1987, A Semantical Approach to Nonmonotonic Logics. In: M. L. Ginsberg (ed.): Readings in Non-Monotonic Reasoning. Los Altos, CA: Morgan Kaufmann, pp Straßer, C.: 2014, Adaptive Logic and Defeasible Reasoning. Applications in Argumentation, Normative Reasoning and Default Reasoning., Vol. 38 of Trends in Logic. Springer. Van De Putte, F. and C. Straßer: 2012, Extending the Standard Format of Adaptive Logics to the Prioritized Case. Logique at Analyse 55(220), Christian Straßer (RUB, UGENT) Tutorial: Nonmonotonic Logic (Day 2) September 3, / 96

Outline. 1 Plausible Reasoning. 2 Preferential / Selection Semantics (KLM, Shoham) 3 Bibliography

Outline. 1 Plausible Reasoning. 2 Preferential / Selection Semantics (KLM, Shoham) 3 Bibliography Outline Tutorial: Nonmonotonic Logic (Day 2) 1 Plausible Reasoning Christian Straßer Institute for Philosophy II, Ruhr-University Bochum Center for Logic and Philosophy of Science, Ghent University http://homepage.ruhr-uni-bochum.de/defeasible-reasoning/index.html

More information

Tutorial: Nonmonotonic Logic

Tutorial: Nonmonotonic Logic Tutorial: Nonmonotonic Logic PhDs in Logic (2017) Christian Straßer May 2, 2017 Outline Defeasible Reasoning Scratching the Surface of Nonmonotonic Logic 1/52 Defeasible Reasoning What is defeasible reasoning?

More information

Tutorial: Nonmonotonic Logic (Day 1)

Tutorial: Nonmonotonic Logic (Day 1) Tutorial: Nonmonotonic Logic (Day 1) Christian Straßer Institute for Philosophy II, Ruhr-University Bochum Center for Logic and Philosophy of Science, Ghent University http://homepage.ruhr-uni-bochum.de/defeasible-reasoning/index.html

More information

On the Semantics of Simple Contrapositive Assumption-Based Argumentation Frameworks

On the Semantics of Simple Contrapositive Assumption-Based Argumentation Frameworks On the Semantics of Simple Contrapositive Assumption-Based Argumentation Frameworks Jesse Heyninck 1 and Ofer Arieli 2 Institute of Philosophy II, Ruhr University Bochum, Germany School of Computer Science,

More information

Belief revision: A vade-mecum

Belief revision: A vade-mecum Belief revision: A vade-mecum Peter Gärdenfors Lund University Cognitive Science, Kungshuset, Lundagård, S 223 50 LUND, Sweden Abstract. This paper contains a brief survey of the area of belief revision

More information

Outline. Adaptive Logics. Introductory Remarks (2) Introductory Remarks (1) Incomplete Survey. Introductory Remarks (3)

Outline. Adaptive Logics. Introductory Remarks (2) Introductory Remarks (1) Incomplete Survey. Introductory Remarks (3) Outline Adaptive Logics The Logics You Always Wanted Diderik Batens Centre for Logic and Philosophy of Science Ghent University, Belgium Diderik.Batens@UGent.be http://logica.ugent.be/dirk/ http://logica.ugent.be/centrum/

More information

On the Semantics of Simple Contrapositive Assumption-Based Argumentation Frameworks

On the Semantics of Simple Contrapositive Assumption-Based Argumentation Frameworks On the Semantics of Simple Contrapositive Assumption-Based Argumentation Frameworks Jesse HEYNINCK a,1 and Ofer ARIELI b,2 a Institute of Philosophy II, Ruhr University Bochum, Germany b School of Computer

More information

4 ENTER Adaptive Logics

4 ENTER Adaptive Logics 4 0 4 ENTER Adaptive Logics 4.1 The problem 4.2 Characterization of an adaptive Logic 4.3 Annotated dynamic proofs: Reliability 4.4 Semantics 4.5 Annotated dynamic proofs: Minimal Abnormality 4.6 Some

More information

Non-monotonic Logic I

Non-monotonic Logic I Non-monotonic Logic I Bridges between classical and non-monotonic consequences Michal Peliš 1 Common reasoning monotonicity Γ ϕ Γ ϕ can fail caused by: background knowledge, implicit facts, presuppositions,

More information

Argumentation-Based Models of Agent Reasoning and Communication

Argumentation-Based Models of Agent Reasoning and Communication Argumentation-Based Models of Agent Reasoning and Communication Sanjay Modgil Department of Informatics, King s College London Outline Logic and Argumentation - Dung s Theory of Argumentation - The Added

More information

Argumentative Characterisations of Non-monotonic Inference in Preferred Subtheories: Stable Equals Preferred

Argumentative Characterisations of Non-monotonic Inference in Preferred Subtheories: Stable Equals Preferred Argumentative Characterisations of Non-monotonic Inference in Preferred Subtheories: Stable Equals Preferred Sanjay Modgil November 17, 2017 Abstract A number of argumentation formalisms provide dialectical

More information

ESSENCE 2014: Argumentation-Based Models of Agent Reasoning and Communication

ESSENCE 2014: Argumentation-Based Models of Agent Reasoning and Communication ESSENCE 2014: Argumentation-Based Models of Agent Reasoning and Communication Sanjay Modgil Department of Informatics, King s College London Outline Logic, Argumentation and Reasoning - Dung s Theory of

More information

A Tableaux Calculus for KLM Preferential and Cumulative Logics

A Tableaux Calculus for KLM Preferential and Cumulative Logics A Tableaux Calculus for KLM Preferential and Cumulative Logics Laura Giordano, Valentina Gliozzi, Nicola Olivetti, Gian Luca Pozzato Dipartimento di Informatica - Università del Piemonte Orientale A. Avogadro

More information

General Patterns for Nonmonotonic Reasoning: From Basic Entailments to Plausible Relations

General Patterns for Nonmonotonic Reasoning: From Basic Entailments to Plausible Relations General Patterns for Nonmonotonic Reasoning: From Basic Entailments to Plausible Relations OFER ARIELI AND ARNON AVRON, Department of Computer Science, School of Mathematical Sciences, Tel-Aviv University,

More information

Analytic Tableau Calculi for KLM Rational Logic R

Analytic Tableau Calculi for KLM Rational Logic R Analytic Tableau Calculi for KLM Rational Logic R Laura Giordano 1, Valentina Gliozzi 2, Nicola Olivetti 3, and Gian Luca Pozzato 2 1 Dipartimento di Informatica - Università del Piemonte Orientale A.

More information

A Lexicographic Inference for Partially Preordered Belief Bases

A Lexicographic Inference for Partially Preordered Belief Bases Proceedings, Eleventh International Conference on Principles of Knowledge Representation and Reasoning (2008) A Lexicographic Inference for Partially Preordered Belief Bases S. Yahi and S. Benferhat and

More information

Argumentation and rules with exceptions

Argumentation and rules with exceptions Argumentation and rules with exceptions Bart VERHEIJ Artificial Intelligence, University of Groningen Abstract. Models of argumentation often take a given set of rules or conditionals as a starting point.

More information

A Universal Logic Approach to Adaptive Logics

A Universal Logic Approach to Adaptive Logics A Universal Logic Approach to Adaptive Logics Diderik Batens Centre for Logic and Philosophy of Science Ghent University, Belgium Diderik.Batens@UGent.be January 11, 2006 Abstract In this paper, adaptive

More information

Handout Lecture 8: Non-monotonic logics

Handout Lecture 8: Non-monotonic logics Handout Lecture 8: Non-monotonic logics Xavier Parent and Leon van der Torre University of Luxembourg April 27, 2016 Abstract This handout is devoted to non-monotonic logics a family of logics devised

More information

Encoding formulas with partially constrained weights in a possibilistic-like many-sorted propositional logic

Encoding formulas with partially constrained weights in a possibilistic-like many-sorted propositional logic Encoding formulas with partially constrained weights in a possibilistic-like many-sorted propositional logic Salem Benferhat CRIL-CNRS, Université d Artois rue Jean Souvraz 62307 Lens Cedex France benferhat@criluniv-artoisfr

More information

Combining Inductive Generalization and Factual Abduction

Combining Inductive Generalization and Factual Abduction Combining Inductive Generalization and Factual Abduction Mathieu Beirlaen Ruhr University Bochum Heinrich Heine University Düsseldorf mathieubeirlaen@gmail.com Abstract. The aim of this paper is to outline

More information

A Propositional Typicality Logic for Extending Rational Consequence

A Propositional Typicality Logic for Extending Rational Consequence A Propositional Typicality Logic for Extending Rational Consequence Richard Booth, Thomas Meyer, Ivan Varzinczak abstract. We introduce Propositional Typicality Logic (PTL), a logic for reasoning about

More information

Kybernetika. Niki Pfeifer; Gernot D. Kleiter Inference in conditional probability logic. Terms of use: Persistent URL:

Kybernetika. Niki Pfeifer; Gernot D. Kleiter Inference in conditional probability logic. Terms of use: Persistent URL: Kybernetika Niki Pfeifer; Gernot D. Kleiter Inference in conditional probability logic Kybernetika, Vol. 42 (2006), No. 4, 391--404 Persistent URL: http://dml.cz/dmlcz/135723 Terms of use: Institute of

More information

Relevance in Structured Argumentation

Relevance in Structured Argumentation Relevance in Structured Argumentation AnneMarie Borg and Christian Straßer, Ruhr-University Bochum, Germany annemarie.borg@rub.de, christian.strasser@rub.de Abstract We study properties related to relevance

More information

Reasoning by Cases in Structured Argumentation.

Reasoning by Cases in Structured Argumentation. . Jesse Heyninck, Mathieu Beirlaen and Christian Straßer Workgroup for Non-Monotonic Logics and Formal Argumentation Institute for Philosophy II Ruhr University Bochum The 32nd ACM SIGAPP Symposium On

More information

Revisiting Unrestricted Rebut and Preferences in Structured Argumentation.

Revisiting Unrestricted Rebut and Preferences in Structured Argumentation. Revisiting Unrestricted Rebut and Preferences in Structured Argumentation. Jesse Heyninck and Christian Straßer Ruhr University Bochum, Germany jesse.heyninck@rub.de, christian.strasser@rub.de Abstract

More information

A Semantic Approach for Iterated Revision in Possibilistic Logic

A Semantic Approach for Iterated Revision in Possibilistic Logic Proceedings of the Twenty-Third AAAI Conference on Artificial Intelligence (2008) A Semantic Approach for Iterated Revision in Possibilistic Logic Guilin Qi AIFB Universität Karlsruhe D-76128 Karlsruhe,

More information

Tackling Defeasible Reasoning in Bochum:

Tackling Defeasible Reasoning in Bochum: Tackling Defeasible Reasoning in Bochum: the Research Group for Non-Monotonic Logic and Formal Argumentation Christian Straßer and Dunja Šešelja April 10, 2017 Outline The NMLFA Reasoning by Cases Unrestricted

More information

Adaptive Logics. p p. Manuel Bremer Centre for Logic, Language and Information

Adaptive Logics. p p. Manuel Bremer Centre for Logic, Language and Information Adaptive Logics The Adaptive Logics program is the third major group in current paraconsistent logic research. Its centre is the group of Diderik Batens at the university of Ghent (forming the "Ghent Centre

More information

Reasoning with Inconsistent and Uncertain Ontologies

Reasoning with Inconsistent and Uncertain Ontologies Reasoning with Inconsistent and Uncertain Ontologies Guilin Qi Southeast University China gqi@seu.edu.cn Reasoning Web 2012 September 05, 2012 Outline Probabilistic logic vs possibilistic logic Probabilistic

More information

Production Inference, Nonmonotonicity and Abduction

Production Inference, Nonmonotonicity and Abduction Production Inference, Nonmonotonicity and Abduction Alexander Bochman Computer Science Department, Holon Academic Institute of Technology, Israel e-mail: bochmana@hait.ac.il Abstract We introduce a general

More information

ESSLLI 2007 COURSE READER. ESSLLI is the Annual Summer School of FoLLI, The Association for Logic, Language and Information

ESSLLI 2007 COURSE READER. ESSLLI is the Annual Summer School of FoLLI, The Association for Logic, Language and Information ESSLLI 2007 19th European Summer School in Logic, Language and Information August 6-17, 2007 http://www.cs.tcd.ie/esslli2007 Trinity College Dublin Ireland COURSE READER ESSLLI is the Annual Summer School

More information

A Unifying Semantics for Belief Change

A Unifying Semantics for Belief Change A Unifying Semantics for Belief Change C0300 Abstract. Many belief change formalisms employ plausibility orderings over the set of possible worlds to determine how the beliefs of an agent ought to be modified

More information

On the Compilation of Stratified Belief Bases under Linear and Possibilistic Logic Policies

On the Compilation of Stratified Belief Bases under Linear and Possibilistic Logic Policies Salem Benferhat CRIL-CNRS, Université d Artois rue Jean Souvraz 62307 Lens Cedex. France. benferhat@cril.univ-artois.fr On the Compilation of Stratified Belief Bases under Linear and Possibilistic Logic

More information

A Discrete Duality Between Nonmonotonic Consequence Relations and Convex Geometries

A Discrete Duality Between Nonmonotonic Consequence Relations and Convex Geometries A Discrete Duality Between Nonmonotonic Consequence Relations and Convex Geometries Johannes Marti and Riccardo Pinosio Draft from April 5, 2018 Abstract In this paper we present a duality between nonmonotonic

More information

Conflict-Based Belief Revision Operators in Possibilistic Logic

Conflict-Based Belief Revision Operators in Possibilistic Logic Conflict-Based Belief Revision Operators in Possibilistic Logic Author Qi, Guilin, Wang, Kewen Published 2012 Conference Title Proceedings of the Twenty-Sixth AAAI Conference on Artificial Intelligence

More information

Human interpretation and reasoning about conditionals

Human interpretation and reasoning about conditionals Human interpretation and reasoning about conditionals Niki Pfeifer 1 Munich Center for Mathematical Philosophy Language and Cognition Ludwig-Maximilians-Universität München www.users.sbg.ac.at/~pfeifern/

More information

The logical meaning of Expansion

The logical meaning of Expansion The logical meaning of Expansion arxiv:cs/0202033v1 [cs.ai] 20 Feb 2002 Daniel Lehmann Institute of Computer Science, Hebrew University, Jerusalem 91904, Israel lehmann@cs.huji.ac.il August 6th, 1999 Abstract

More information

A modal perspective on defeasible reasoning

A modal perspective on defeasible reasoning A modal perspective on defeasible reasoning K. Britz 1, J. Heidema and W.A. Labuschagne abstract. We introduce various new supraclassical entailment relations for defeasible reasoning and investigate some

More information

The Logic of Confirmation and Theory Assessment

The Logic of Confirmation and Theory Assessment The Logic of Confirmation and Theory Assessment The Logic of Confirmation and Theory Assessment 161 Franz Huber 1. Hempel s conditions of adequacy In his Studies in the Logic of Confirmation (1945) Carl

More information

Simulating Human Inferences in the Light of New Information: A Formal Analysis

Simulating Human Inferences in the Light of New Information: A Formal Analysis Proceedings of the Twenty-Fifth International Joint Conference on Artificial Intelligence (IJCAI-16) Simulating Human Inferences in the Light of New Information: A Formal Analysis Marco Ragni, 1, Christian

More information

An adaptive logic for relevant classical deduction

An adaptive logic for relevant classical deduction Journal of Applied Logic 5 (2007) 602 612 www.elsevier.com/locate/jal An adaptive logic for relevant classical deduction Hans Lycke Centre for Logic and Philosophy of Science, Universiteit Gent, Belgium

More information

Default Reasoning and Belief Revision: A Syntax-Independent Approach. (Extended Abstract) Department of Computer Science and Engineering

Default Reasoning and Belief Revision: A Syntax-Independent Approach. (Extended Abstract) Department of Computer Science and Engineering Default Reasoning and Belief Revision: A Syntax-Independent Approach (Extended Abstract) Dongmo Zhang 1;2, Zhaohui Zhu 1 and Shifu Chen 2 1 Department of Computer Science and Engineering Nanjing University

More information

Nonmonotonic Modes of Inference

Nonmonotonic Modes of Inference Proceedings of the Twenty-Third AAAI Conference on Artificial Intelligence (2008) Nonmonotonic Modes of Inference Victor Jauregui The University of New South Wales Sydney, Australia. Abstract In this paper

More information

Journal of Applied Logic

Journal of Applied Logic Journal of Applied Logic 11 (2013) 147 168 Contents lists available at SciVerse ScienceDirect Journal of Applied Logic www.elsevier.com/locate/jal Two adaptive logics of norm-propositions Mathieu Beirlaen,

More information

A LOGIC WITH BIG-STEPPED PROBABILITIES THAT CAN MODEL NONMONOTONIC REASONING OF SYSTEM P. Dragan Doder

A LOGIC WITH BIG-STEPPED PROBABILITIES THAT CAN MODEL NONMONOTONIC REASONING OF SYSTEM P. Dragan Doder PUBLICATIONS DE L INSTITUT MATHÉMATIQUE Nouvelle série, tome 90(104) (2011), 13 22 DOI: 10.2298/PIM1104013D A LOGIC WITH BIG-STEPPED PROBABILITIES THAT CAN MODEL NONMONOTONIC REASONING OF SYSTEM P Dragan

More information

Reasoning: From Basic Entailments. to Plausible Relations. Department of Computer Science. School of Mathematical Sciences. Tel-Aviv University

Reasoning: From Basic Entailments. to Plausible Relations. Department of Computer Science. School of Mathematical Sciences. Tel-Aviv University General Patterns for Nonmonotonic Reasoning: From Basic Entailments to Plausible Relations Ofer Arieli Arnon Avron Department of Computer Science School of Mathematical Sciences Tel-Aviv University Tel-Aviv

More information

Prioritized Sequent-Based Argumentation

Prioritized Sequent-Based Argumentation Prioritized Sequent-Based Argumentation Ofer Arieli School of Computer Science Tel-Aviv Academic College, Israel oarieli@mta.ac.il AnneMarie Borg Institute of Philosophy II Ruhr-University Bochum, Germany

More information

The Logic of Theory Assessment*

The Logic of Theory Assessment* The Logic of Theory Assessment* Franz Huber, California Institute of Technology penultimate version: please cite the paper in the Journal of Philosophical Logic Contents 1 Hempel s Logic of Confirmation

More information

Concept Model Semantics for DL Preferential Reasoning

Concept Model Semantics for DL Preferential Reasoning Concept Model Semantics for DL Preferential Reasoning Katarina Britz 1,2, Thomas Meyer 1,2, and Ivan Varzinczak 1,2 1 CSIR Meraka Institute, Pretoria, South Africa 2 University of KwaZulu-Natal, Durban,

More information

Metatheory of The Standard Format

Metatheory of The Standard Format Chapter 5 Metatheory of The Standard Format In this chapter many properties of adaptive logics in standard format are stated and proven. All theorems proven apply to all the adaptive logics. That is the

More information

An Egalitarist Fusion of Incommensurable Ranked Belief Bases under Constraints

An Egalitarist Fusion of Incommensurable Ranked Belief Bases under Constraints An Egalitarist Fusion of Incommensurable Ranked Belief Bases under Constraints Salem Benferhat and Sylvain Lagrue and Julien Rossit CRIL - Université d Artois Faculté des Sciences Jean Perrin Rue Jean

More information

Nested Epistemic Logic Programs

Nested Epistemic Logic Programs Nested Epistemic Logic Programs Kewen Wang 1 and Yan Zhang 2 1 Griffith University, Australia k.wang@griffith.edu.au 2 University of Western Sydney yan@cit.uws.edu.au Abstract. Nested logic programs and

More information

Adaptively Applying Modus Ponens in Conditional Logics of Normality

Adaptively Applying Modus Ponens in Conditional Logics of Normality Adaptively Applying Modus Ponens in Conditional Logics of Normality Christian Straßer Centre for Logic and Philosophy of Science, Ghent University (UGent) Blandijnberg 2, 9000 Gent, Belgium Email: Ö Ø

More information

Conditional Logic and Belief Revision

Conditional Logic and Belief Revision Conditional Logic and Belief Revision Ginger Schultheis (vks@mit.edu) and David Boylan (dboylan@mit.edu) January 2017 History The formal study of belief revision grew out out of two research traditions:

More information

Possibilistic Safe Beliefs

Possibilistic Safe Beliefs Possibilistic Safe Beliefs Oscar Estrada 1, José Arrazola 1, and Mauricio Osorio 2 1 Benemérita Universidad Autónoma de Puebla oestrada2005@gmail.com, arrazola@fcfm.buap.mx. 2 Universidad de las Américas

More information

Ordinal and Probabilistic Representations of Acceptance

Ordinal and Probabilistic Representations of Acceptance Journal of Artificial Intelligence Research 22 (2004) 23-56 Submitted 05/03; published 07/04 Ordinal and Probabilistic Representations of Acceptance Didier Dubois Helene Fargier Henri Prade Institut de

More information

EQUIVALENCE OF THE INFORMATION STRUCTURE WITH UNAWARENESS TO THE LOGIC OF AWARENESS. 1. Introduction

EQUIVALENCE OF THE INFORMATION STRUCTURE WITH UNAWARENESS TO THE LOGIC OF AWARENESS. 1. Introduction EQUIVALENCE OF THE INFORMATION STRUCTURE WITH UNAWARENESS TO THE LOGIC OF AWARENESS SANDER HEINSALU Abstract. Here it is shown that the unawareness structure in Li (29) is equivalent to a single-agent

More information

Conditional Probability and Defeasible Inference

Conditional Probability and Defeasible Inference Carnegie Mellon University Research Showcase @ CMU Department of Philosophy Dietrich College of Humanities and Social Sciences 2004 Conditional Probability and Defeasible Inference Horacio Arlo Costa Carnegie

More information

Framing human inference by coherence based probability logic

Framing human inference by coherence based probability logic Framing human inference by coherence based probability logic Pfeifer, N., & Kleiter, G. D. University of Salzburg (Austria) p. 1 Why coherence based? Degrees of belief & subjective probabilities versus

More information

Bisimulation for conditional modalities

Bisimulation for conditional modalities Bisimulation for conditional modalities Alexandru Baltag and Giovanni Ciná Institute for Logic, Language and Computation, University of Amsterdam March 21, 2016 Abstract We give a general definition of

More information

Applied Logic. Lecture 1 - Propositional logic. Marcin Szczuka. Institute of Informatics, The University of Warsaw

Applied Logic. Lecture 1 - Propositional logic. Marcin Szczuka. Institute of Informatics, The University of Warsaw Applied Logic Lecture 1 - Propositional logic Marcin Szczuka Institute of Informatics, The University of Warsaw Monographic lecture, Spring semester 2017/2018 Marcin Szczuka (MIMUW) Applied Logic 2018

More information

Inductive, Abductive and Pragmatic Reasoning

Inductive, Abductive and Pragmatic Reasoning Inductive, Abductive and Pragmatic Reasoning Abstract This paper gives a modern version of Pierce s distinction between induction and abduction, according to which they are both forms of pragmatic (or

More information

A Preference Logic With Four Kinds of Preferences

A Preference Logic With Four Kinds of Preferences A Preference Logic With Four Kinds of Preferences Zhang Zhizheng and Xing Hancheng School of Computer Science and Engineering, Southeast University No.2 Sipailou, Nanjing, China {seu_zzz; xhc}@seu.edu.cn

More information

Postulates for logic-based argumentation systems

Postulates for logic-based argumentation systems Postulates for logic-based argumentation systems Leila Amgoud IRIT CNRS Toulouse France Abstract Logic-based argumentation systems are developed for reasoning with inconsistent information. Starting from

More information

COMP 182 Algorithmic Thinking. Proofs. Luay Nakhleh Computer Science Rice University

COMP 182 Algorithmic Thinking. Proofs. Luay Nakhleh Computer Science Rice University COMP 182 Algorithmic Thinking Proofs Luay Nakhleh Computer Science Rice University 1 Reading Material Chapter 1, Section 3, 6, 7, 8 Propositional Equivalences The compound propositions p and q are called

More information

Argument-based Expansion Operators in Possibilistic Defeasible Logic Programming: Characterization and Logical Properties

Argument-based Expansion Operators in Possibilistic Defeasible Logic Programming: Characterization and Logical Properties Argument-based Expansion Operators in Possibilistic Defeasible Logic Programming: Characterization and Logical Properties 2 Carlos I. Chesñevar 1, Guillermo R. Simari 2, Lluis Godo 3, and Teresa Alsinet

More information

Plausibility structures for default reasoning

Plausibility structures for default reasoning Plausibility structures for default reasoning Yves Moinard 1 Abstract. Friedman and Halpern have introduced the inference by plausibility structures, which provides semantics for various default logics.

More information

Reconstructing an Agent s Epistemic State from Observations

Reconstructing an Agent s Epistemic State from Observations Reconstructing an Agent s Epistemic State from Observations Richard Booth Macquarie University Dept. of Computing Sydney NSW 2109 Australia rbooth@ics.mq.edu.au Alexander Nittka University of Leipzig Dept.

More information

Technical R e p o r t. Merging in the Horn Fragment DBAI-TR Adrian Haret, Stefan Rümmele, Stefan Woltran. Artificial Intelligence

Technical R e p o r t. Merging in the Horn Fragment DBAI-TR Adrian Haret, Stefan Rümmele, Stefan Woltran. Artificial Intelligence Technical R e p o r t Institut für Informationssysteme Abteilung Datenbanken und Artificial Intelligence Merging in the Horn Fragment DBAI-TR-2015-91 Adrian Haret, Stefan Rümmele, Stefan Woltran Institut

More information

Characterization of Semantics for Argument Systems

Characterization of Semantics for Argument Systems Characterization of Semantics for Argument Systems Philippe Besnard and Sylvie Doutre IRIT Université Paul Sabatier 118, route de Narbonne 31062 Toulouse Cedex 4 France besnard, doutre}@irit.fr Abstract

More information

Which Style of Reasoning to Choose in the Face of Conflicting Information?

Which Style of Reasoning to Choose in the Face of Conflicting Information? Which Style of Reasoning to Choose in the Face of Conflicting Information? Joke Meheus Christian Straßer Peter Verdée Centre for Logic and Philosophy of Science University of Ghent, Belgium {Joke.Meheus,Christian.Strasser,Peter.Verdee}@UGent.be

More information

An argumentation system for reasoning with conflict-minimal paraconsistent ALC

An argumentation system for reasoning with conflict-minimal paraconsistent ALC An argumentation system for reasoning with conflict-minimal paraconsistent ALC Wenzhao Qiao and Nico Roos Department of Knowledge Engineering, Maastricht University Bouillonstraat 8-10, 6211 LH Maastricht,

More information

A Possibilistic Extension of Description Logics

A Possibilistic Extension of Description Logics A Possibilistic Extension of Description Logics Guilin Qi 1, Jeff Z. Pan 2, and Qiu Ji 1 1 Institute AIFB, University of Karlsruhe, Germany {gqi,qiji}@aifb.uni-karlsruhe.de 2 Department of Computing Science,

More information

Reasoning under inconsistency: the forgotten connective

Reasoning under inconsistency: the forgotten connective Reasoning under inconsistency: the forgotten connective Sébastien Konieczny CRIL - Université d Artois 62307 Lens, France konieczny@cril.univ-artois.fr Jérôme Lang IRIT - Université Paul Sabatier 31062

More information

Revising Nonmonotonic Theories: The Case of Defeasible Logic

Revising Nonmonotonic Theories: The Case of Defeasible Logic Revising Nonmonotonic Theories: The Case of Defeasible Logic D. Billington, G. Antoniou, G. Governatori, and M. Maher School of Computing and Information Technology Griffith University, QLD 4111, Australia

More information

A DISJUNCTION IS EXCLUSIVE UNTIL PROVEN OTHERWISE INTRODUCING THE ADAPTIVE LOGICS APPROACH TO GRICEAN PRAGMATICS

A DISJUNCTION IS EXCLUSIVE UNTIL PROVEN OTHERWISE INTRODUCING THE ADAPTIVE LOGICS APPROACH TO GRICEAN PRAGMATICS A DISJUNCTION IS EXCLUSIVE UNTIL PROVEN OTHERWISE INTRODUCING THE ADAPTIVE LOGICS APPROACH TO GRICEAN PRAGMATICS Hans Lycke Abstract In Gricean pragmatics, generalized conversational implicatures (GCI)

More information

Outline. Golden Rule

Outline. Golden Rule Outline Christian Straßer Institute for Philosophy II, Ruhr-University Bochum Center for Logic and Philosophy of Science, Ghent University http://homepage.ruhr-uni-bochum.de/defeasible-reasoning/index.html

More information

Belief functions and default reasoning

Belief functions and default reasoning Artificial Intelligence 122 (2000) 1 69 Belief functions and default reasoning S. Benferhat a,,a.saffiotti b,p.smets c a IRIT, Université Paul Sabatier, 118 Route de Narbonne, Toulouse 31062 Cedex, France

More information

Adaptive Logic Characterizations of Input/Output Logic

Adaptive Logic Characterizations of Input/Output Logic Christian Straßer Mathieu Beirlaen Frederik Van De Putte Adaptive Logic Characterizations of Input/Output Logic Abstract. We translate unconstrained and constrained input/output logics as introduced by

More information

Towards Conditional Logic Semantics for Abstract Dialectical Frameworks

Towards Conditional Logic Semantics for Abstract Dialectical Frameworks Towards Conditional Logic Semantics for Abstract Dialectical Frameworks Gabriele Kern-Isberner 1 and Matthias Thimm 2 1 Department of Computer Science, Technical University Dortmund, Germany 2 Institute

More information

Quasi-merging and Pure-arbitration on Information for the Family of Adaptive Logics ADM

Quasi-merging and Pure-arbitration on Information for the Family of Adaptive Logics ADM Quasi-merging and Pure-arbitration on Information for the Family of Adaptive Logics ADM Giuseppe Primiero Giuseppe.Primiero@UGent.be Centre for Logic and Philosophy of Science, Ghent University, Belgium

More information

A Split-combination Method for Merging Inconsistent Possibilistic Knowledge Bases

A Split-combination Method for Merging Inconsistent Possibilistic Knowledge Bases A Split-combination Method for Merging Inconsistent Possibilistic Knowledge Bases Guilin Qi 1,2, Weiru Liu 1, David H. Glass 2 1 School of Computer Science Queen s University Belfast, Belfast BT7 1NN 2

More information

On iterated revision in the AGM framework

On iterated revision in the AGM framework On iterated revision in the AGM framework Andreas Herzig, Sébastien Konieczny, Laurent Perrussel Institut de Recherche en Informatique de Toulouse 118 route de Narbonne - 31062 Toulouse - France {herzig,konieczny,perrussel}@irit.fr

More information

Using Defeasible Information to Obtain Coherence

Using Defeasible Information to Obtain Coherence Using Defeasible Information to Obtain Coherence Giovanni Casini Thomas Meyer March 26, 2017 Abstract We consider the problem of obtaining coherence in a propositional knowledge base using techniques from

More information

Formal Epistemology: Lecture Notes. Horacio Arló-Costa Carnegie Mellon University

Formal Epistemology: Lecture Notes. Horacio Arló-Costa Carnegie Mellon University Formal Epistemology: Lecture Notes Horacio Arló-Costa Carnegie Mellon University hcosta@andrew.cmu.edu Bayesian Epistemology Radical probabilism doesn t insists that probabilities be based on certainties;

More information

Chapter 2 Background. 2.1 A Basic Description Logic

Chapter 2 Background. 2.1 A Basic Description Logic Chapter 2 Background Abstract Description Logics is a family of knowledge representation formalisms used to represent knowledge of a domain, usually called world. For that, it first defines the relevant

More information

Comment on Leitgeb s Stability Theory of Belief

Comment on Leitgeb s Stability Theory of Belief Comment on Leitgeb s Stability Theory of Belief Hanti Lin Kevin T. Kelly Carnegie Mellon University {hantil, kk3n}@andrew.cmu.edu Hannes Leitgeb s stability theory of belief provides three synchronic constraints

More information

Logic for Computer Science - Week 4 Natural Deduction

Logic for Computer Science - Week 4 Natural Deduction Logic for Computer Science - Week 4 Natural Deduction 1 Introduction In the previous lecture we have discussed some important notions about the semantics of propositional logic. 1. the truth value of a

More information

On Warranted Inference in Possibilistic Defeasible Logic Programming 1

On Warranted Inference in Possibilistic Defeasible Logic Programming 1 On Warranted Inference in Possibilistic Defeasible Logic Programming 1 Carlos Chesñevar a,2, Guillermo Simari b Lluís Godo c and Teresa Alsinet a a Department of Computer Science. University of Lleida,

More information

From Causal Models To Counterfactual Structures

From Causal Models To Counterfactual Structures From Causal Models To Counterfactual Structures Joseph Y. Halpern Cornell University halpern@cs.cornell.edu June 14, 2011 Abstract Galles and Pearl [1998] claimed that for recursive models, the causal

More information

Completeness in the Monadic Predicate Calculus. We have a system of eight rules of proof. Let's list them:

Completeness in the Monadic Predicate Calculus. We have a system of eight rules of proof. Let's list them: Completeness in the Monadic Predicate Calculus We have a system of eight rules of proof. Let's list them: PI At any stage of a derivation, you may write down a sentence φ with {φ} as its premiss set. TC

More information

Adaptive Logics for Abduction and the Explication of Explanation-Seeking Processes

Adaptive Logics for Abduction and the Explication of Explanation-Seeking Processes Adaptive Logics for Abduction and the Explication of Explanation-Seeking Processes Joke Meheus Centre for Logic and Philosophy of Science Ghent University Blandijnberg 2 9000 Ghent, Belgium tel: ++ 32

More information

Merging Stratified Knowledge Bases under Constraints

Merging Stratified Knowledge Bases under Constraints Merging Stratified Knowledge Bases under Constraints Guilin Qi, Weiru Liu, David A. Bell School of Electronics, Electrical Engineering and Computer Science Queen s University Belfast Belfast, BT7 1NN,

More information

arxiv: v1 [math.pr] 20 Mar 2013

arxiv: v1 [math.pr] 20 Mar 2013 Quasi Conjunction, Quasi Disjunction, T-norms and T-conorms: Probabilistic Aspects Angelo Gilio a, Giuseppe Sanfilippo b, a Dipartimento di Scienze di Base e Applicate per l Ingegneria, University of Rome

More information

A Formal Logic for the Abduction of Singular Hypotheses

A Formal Logic for the Abduction of Singular Hypotheses A Formal Logic for the Abduction of Singular Hypotheses Joke Meheus Centre for Logic and Philosophy of Science University of Ghent, Belgium Joke.Meheus@UGent.be This is the final version of the paper as

More information

Canonical Calculi: Invertibility, Axiom expansion and (Non)-determinism

Canonical Calculi: Invertibility, Axiom expansion and (Non)-determinism Canonical Calculi: Invertibility, Axiom expansion and (Non)-determinism A. Avron 1, A. Ciabattoni 2, and A. Zamansky 1 1 Tel-Aviv University 2 Vienna University of Technology Abstract. We apply the semantic

More information

What is a Default Priority?

What is a Default Priority? What is a Default Priority? Craig Boutilier Department of Computer Science University of British Columbia Vancouver, British Columbia CANADA, V6T 1Z2 email: cebly@cs.ubc.ca Abstract The notion of default

More information

Logic: Propositional Logic Truth Tables

Logic: Propositional Logic Truth Tables Logic: Propositional Logic Truth Tables Raffaella Bernardi bernardi@inf.unibz.it P.zza Domenicani 3, Room 2.28 Faculty of Computer Science, Free University of Bolzano-Bozen http://www.inf.unibz.it/~bernardi/courses/logic06

More information

An Experimental Analysis of Possibilistic Default Reasoning

An Experimental Analysis of Possibilistic Default Reasoning An Experimental Analysis of Possibilistic Default Reasoning Salem Benferhat 1 Jean François Bonnefon 2 Rui Da Silva Neves 2 1 CRIL-CNRS, Université d Artois 2 D.S.V.P., Université Toulouse Le Mirail Faculté

More information