Lecture 8. FOL
SlidesInfo
Here you can find the lecture slides.
You can use them to follow along during the lecture and to go through the material again after.
Feel free to ask questions during the lecture!!
Technical info
You can use the arrow keys to navigate through the slides, or alternatively, click the little arrows on the slides themselves.
Pressing the F-key while the slides are selected, puts them in fullscreen. You can quit fullscreen by pressing ESC.
There are more the slides can do, find out under https://revealjs.com/ .
∀I
Logical methods for AI
Lecture 8
FOL
This work is licensed under CC BY 4.0
-
First Order Logic
Background
- Logical system for valid inference with quantifiers:
- "For all"
- "There is"
- Benchmark system: strong!
- Theoretical limitations
-
FOL syntax
Alphabet
- Non-logical symbols:
- Individual constants: $a,b,c,\dots,$ also $Alan$ or $Ada$.
- Function symbols: $f,g,h,\dots,$ also $MotherOf$ or $BirthplaceOf$.
- Predicates: $P,Q,R,\dots,$ also $Blue$ or $BiggerThan$.
Alphabet
- Logical symbols:
- Connectives: $\neg,\land,\lor,\to,\leftrightarrow$
- Variables: $x,y,z,\dots$
- Quantifiers: $\forall,\exists$
- Auxilliaries: $(,),\ ","$
Terms
$$\langle const\rangle ::= a \mid b\mid \dots $$ $$\langle var\rangle ::= x \mid y\mid \dots $$ $$\langle fun^n\rangle ::= f^n\mid g^n\mid \dots $$ $$\langle term\rangle::= \langle const\rangle\mid\langle variable\rangle\mid \langle fun^n\rangle(\underbrace{\langle term\rangle,\dots,\langle term\rangle}_{n\text{ times}})$$Terms
Denote individuals:
- Constants: Proper names, like "Utrecht", "Johannes", ...
- Variables: Pronouns: "he", "she", "it", ...
- Functions: Expressions like "the father of", "the birthplace of"
Atomic formulas
$$\langle pred^n\rangle ::= P^n\mid Q^n\mid \dots $$ $$\langle atom\rangle::= \langle pred^n\rangle(\underbrace{\langle term\rangle,\dots\langle term\rangle}_{n\text{ times}})$$Atomic formulas
- Simple, term-predicate sentences: the robot is smart
- Predicate: "... is smart", "... is bigger than ..."
- Examples:
- $\mathbf{Smart}(Robot)$
- $\mathbf{LLM}(x)$
- $\mathbf{BiggerThan}(Robot,Johannes)$
- $\mathsf{Lord}(\mathsf{FatherOf}(Ada))$
Quantifiers
$$\langle var\rangle ::= x \mid y\mid \dots $$ $$\langle quant\rangle ::= \forall\mid\exists$$ $$\langle fml\rangle::=\langle atom\rangle\mid\langle unop\rangle\langle fml\rangle\mid$$ $$(\langle fml\rangle\langle binop\rangle \langle fml\rangle)\mid \langle quant\rangle \langle var\rangle\langle fml\rangle$$Quantifiers
- General claims: $$\forall x\exists y\mathsf{BiggerThan}(x,y)$$
- For everything there is something, such that it is bigger than it.
Identity/equality
$$t_1=t_2 \qquad \qquad t_1\neq t_2$$ $$\mathsf{FatherOf}(Ada)=Byron$$ $$\exists x\exists y(\mathsf{LLM}(x)\land\mathsf{LLM}(y)\land x\neq y)$$ $$\forall x\forall y((\mathsf{GOAT}(x)\land\mathsf{GOAT}(y))\to x=y)$$-
FOL syntax
Domains
- Objects we're talking about.
- $\mathcal{D}=\Set{o_1,o_2,\dots}$
- In KR: representations of the world.
- Often graphs.
Denotations
- Constants: $c^\mathcal{M}\in\mathcal{D}$
- Assignments: $\alpha:\Set{x,y,z,\dots}\to\mathcal{D}$
- Variables: $\alpha(x)\in\mathcal{D}$
- Functions: $f^\mathcal{M}:\mathcal{D}^n\to \mathcal{D}$
- Recursion: $$t^\mathcal{M}_\alpha::= c^\mathcal{M}\mid \alpha(x) \mid f^\mathcal{M}((t_1)^\mathcal{M}_\alpha, \dots (t_n)^\mathcal{M}_\alpha) $$
Extensions
- Set of things having a property/standing in a relation: $$P^\mathcal{M}\subseteq \mathcal{D}^n$$
- $\mathcal{D}^n=\Set{\langle o_1,\dots,o_n\rangle:o_i\in\mathcal{D}}$
- $\mathsf{Sister}^\mathcal{M}=\Set{\langle o_1,o_2\rangle:o_1\text{ is sister of }o_2\text{ in }\mathcal{M}}$
Atomic truth
$$\nu_\alpha(\mathsf{Parent}(\mathsf{FatherOf}(Ada),Clara))=1$$ $$\Leftrightarrow$$ $$\langle\mathsf{FatherOf}^\mathcal{M}(Ada^\mathcal{M}),Clara^\mathcal{M}\rangle\in \mathsf{Parent}^\mathcal{M}$$ $$=\Set{\langle o_1,o_2\rangle:o_1\text{ is parent of }o_2\text{ in }\mathcal{M}}$$Atomic truth
$$\nu_\alpha(\mathsf{Blue}(x))=1\Leftrightarrow \alpha(x)\in\mathsf{Blue}^\mathcal{M}$$Atomic truth
$$\nu_\alpha(P(t_1,\dots,t_n))=1\Leftrightarrow \langle (t_1)^\mathcal{M}_\alpha, \dots, (t_n)^\mathcal{M}_\alpha\rangle\in P^\mathcal{M}$$Identity truth
$$\nu_\alpha(\mathsf{FatherOf}(Ada)=x)=1$$ $$\Leftrightarrow$$ $$\mathsf{FatherOf}^\mathcal{M}(Ada^\mathcal{M})=\alpha(x)$$Identity truth
$$\nu_\alpha(t_1=t_2)=1\Leftrightarrow t_1^\mathcal{M}=t_2^\mathcal{M}$$Universal truth
$$\nu\_\alpha(\forall x\mathsf{Blue}(x))$$ $$=_{\mathbf{o}^\mathcal{M}=o}$$ $$\nu_\alpha(\mathsf{Blue}(\mathbf{o}_1))\times\nu\_\alpha(\mathsf{Blue}(\mathbf{o}_2)\times\dots$$ $$=$$ $$\prod_{o\in\mathcal{D}}\nu_\alpha(\mathsf{Blue}(\mathbf{o}))$$Existential truth
$$\nu\_\alpha(\exists x\mathsf{Blue}(x))$$ $$=_{\mathbf{o}^\mathcal{M}=o}$$ $$\nu_\alpha(\mathsf{Blue}(\mathbf{o}_1))+\nu\_\alpha(\mathsf{Blue}(\mathbf{o}_2)+\dots$$ $$=$$ $$\sum_{o\in\mathcal{D}}\nu_\alpha(\mathsf{Blue}(\mathbf{o}))$$FOL consequence
$$P_1,P_2,\dots \vDash C$$ $$\Leftrightarrow$$ $$\text{ for all }\mathcal{M},\nu,\alpha,$$ $$\text{ if }\nu_\alpha(P_1)=1,\nu_\alpha(P_2)=1,\dots\text{, then }$$ $$\nu_\alpha(C)=1.$$Quantifier laws
Duality $\exists/\forall$ | $\exists xA\vDash \neg\forall x\neg A\qquad \neg\forall x\neg A\vDash \exists xA$ | |
Duality $\forall/\exists$ | $\forall xA\vDash \neg\exists x\neg A\qquad \neg\exists x\neg A\vDash \forall xA$ |
Quantifier laws
Distr. $\forall/\land$ | $\forall x(A\land B)\vDash \forall xA\land \forall xB\quad \forall x A\land \forall x B\vDash \forall x(A\land B)$ | |
Distr. $\exists/\lor$ | $\exists x(A\lor B)\vDash \exists xA\lor \exists xB\quad \exists x A\lor \exists x B\vDash \exists x(A\lor B)$ |
Quantifier laws
Commutativity $\exists$ | $\exists x\exists yA\vDash \exists y\exists xA$ | |
Commutativity $\forall$ | $\forall x\forall yA\vDash \forall y\forall xA$ |
Quantifier laws
$\forall/\exists$ | $\exists x\forall y A(x,y)\vDash \forall y\exists x A(x,y)$ |
-
FOL's limitations
Limitations
- Gödel's original completeness theorem:
- Every consistent logical system which models mathematical inference has an undecidable truth.
- No mathematically omniscient KB 😦
Limitations
- Turing's undecidability theorem:
- There's no algorithm for determining whether any given FOL inference is valid.
- No fully automated perfect reasoner 😢