-
Aristotle, Van Benthem, Moss
-
Natural language inference
- No translating into FOL
-
Special-purpose logic engineering
- Complete and decidable logics
- 5 different systems in this talk alone!
- Restricted domain: Reasoning about sizes
- Why not embed into FOL with Arithmetic?
- Decidable
$\to$ More cognitively plausible - Light
$\to$ More cognitively plausible - Counting is not needed:
- “There is more sand than water in the pond”
- Decidable
-
$a, b, c, ...$ are nouns -
$all(a, b)$ -
$some(a, b)$ -
$more(a, b)$ -
$atLeast(a, b)$ -
No connectives $\and, \or \neg$
-
No quantifiers $\forall, \exists$
-
Models:
$\mathcal{M}$ consisting of set$M$ . -
For all nouns
$a$ ,$[![a]!] \sube M$ -
$\mathcal{M} \vDash all(a, b)$ iff$[![a]!] \sube [![b]!]$ -
$\mathcal{M} \vDash some(a, b)$ iff$[![a]!] \cap [![b]!] \ne \empty$ -
$\mathcal{M} \vDash more(a, b)$ iff$|[![a]!]| \gt |[![b]!]|$ -
$\mathcal{M} \vDash atLeast(a, b)$ iff$|[![a]!]| \ge |[![b]!]|$ -
$\Gamma \vDash \varphi$ iff every FINITE model$\mathcal{M}$ which satisfies$\Gamma$ satisfies$\varphi$ .
See board
Theorem:
- From Larry Moss’ “Syllogistic Logic with Cardinality Comparisons”
- For a noun
$a$ , we allow$a$ to be complemented:$\bar a$ -
$\bar{\bar{a}} = a$ at syntax level -
Semantics:
$[![\bar a]!] = M \setminus [![a]!]$ -
Rules of Inference:
- See board
Theorem:
- More complicated than the proof for
$\mathcal{S(card)}$ !
Bonus Theorem:
-
$all(a, b)$ -
$some(a, b)$ -
$most(a, b)$ -
$\mathcal{M} \vDash most(a, b)$ iff$|[![a]!] \cap [![b]!]| \gt \dfrac{1}{2}[![a]!]$
See board
Theorem:
$\big \downarrow$
$\big \downarrow$ $\big \downarrow$
$\big \downarrow$ $\big \downarrow$
-
We want a complete logic involving ‘more than’ and ‘most’
-
Can’t just combine rules, because of interplay between ‘more than’ and ‘most’
-
$$most(a, b) \equiv more(a \cap b, a \cap \bar b)$$ -
So we need noun complement and noun intersection
-
It is easier to make rules for Union
$\cup$ $a \cap b \equiv \overline{\bar a \cap \bar b}$
- For base nouns
$a, b$ , we allow$a$ and$b$ to be unioned:$$a \cup b$$ - We assume laws of boolean algebra at syntax level (?)
-
Semantics:
$$[![a \cup b]!] = [![a]!] \cup [![b]!]$$ -
Rules of Inference:
- See board
Conjecture: The set of rules given above for
- The Holy Grail!
- What is a complete set of rules?
- Can’t just haphazardly combine
$\mathcal{S^{\dagger}(card)}$ and$\mathcal{S^{\cup}(card)}$
- Can’t just haphazardly combine
See board
- We can derive the first 6 ‘most’ rules already!
- Remember that infinite schema!