1. NLP

2. Introduction to NLP
Inference

3. Modus Ponens Modus ponens: Example:   ⇒   Cat(Martin)
 x: Cat(x) ⇒ EatsFish(x)
EatsFish(Martin)

4. Inference Forward chaining Backward chaining Prolog example
as individual facts are added to the database, all derived inferences are generated
Backward chaining
starts from queries
Example: the Prolog programming language
Prolog example
father(X, Y) :- parent(X, Y), male(X). parent(john, bill). parent(jane, bill). female(jane). male (john). ?- father(M, bill).

5. Examples Brothers are siblings One's mother is one's female parent
The kinship domain:
Brothers are siblings
x,y Brother(x,y)  Sibling(x,y)
One's mother is one's female parent
m,c Mother(c) = m  (Female(m)  Parent(m,c))
“Sibling” is symmetric
x,y Sibling(x,y)  Sibling(y,x)

6. Universal Instantiation
Every instantiation of a universally quantified sentence is entailed by it:
v α
Subst({v/g}, α)
for any variable v and ground term g
E.g., x Cat(x)  Fish(y)  Eats(x,y) yields:
Cat(Martin)  Fish(Blub)  Eats(Martin,Blub)

7. Existential Instantiation
For any sentence α, variable v, and constant symbol k that does not appear elsewhere in the knowledge base:
v α
Subst({v/k}, α)
E.g., x Cat(x)  EatsFish(x) yields:
Cat(C1)  EatsFish(C1)
provided C1 is a new constant symbol, called a Skolem constant

8. Unification If a substitution θ is available, unification is possible
Examples:
p=Eats(x,y), q=Eats(x,Blub), possible if θ = {y/Blub}
p=Eats(Martin,y), q=Eats(x,Blub), possible if θ = {x/Martin,y/Blub}
p=Eats(Martin,y), q=Eats(y,Blub), fails because Martin≠Blub
Subsumption
Unification works not only when two things are the same but also when one of them subsumes the other one
Example: All cats eat fish, Martin is a cat, Blub is a fish

9. NLP

10. Introduction to NLP
Semantic Parsing

11. Semantic Parsing Converting natural language to a logical form
e.g., executable code for a specific application
Example:
Airline reservations
Geographical query systems

12. Stages of Semantic Parsing
Input
Sentence
Syntactic Analysis
Syntactic structure
Semantic Analysis
Semantic representation

13. Compositional Semantics
Add semantic attachments to CFG rules
Compositional semantics
Parse the sentence syntactically
Associate some semantics to each word
Combine the semantics of words and non-terminals recursively
Until the root of the sentence

14. Example
Input
Javier likes pizza
Output
like(Javier, pizza)

15. Semantic Parsing Associate a semantic expression with each node
S: likes(Javier, pizza)
N: Javier
VP: λx likes(x,pizza)
Javier
V: λ x,y likes(x,y)
N: pizza
likes
pizza

16. Figure 18.4

17. Using CCG (Steedman 1996) CCG representations for semantics
ADJ: λx.tall(x)
(S\NP)/ADJ : λf.λx.f(x)
NP: YaoMing
YaoMing is tall
NP (S\NP)/ADJ ADJ
YaoMing λf.λx.f(x) λx.tall(x)
>
S\NP
λx.tall(x)
< S
Tall (YaoMing)

18. CCG
NACLO problem from 2014
Authors: Jonathan Kummerfeld, Aleka Blackwell, and Patrick Littell

19. CCG

20. CCG

21. CCG

22. CCG

23. Answer

24. CCG

25. CCG

26. CCG

27. CCG

28. CCG

29. CCG

30. GeoQuery (Zelle and Mooney 1996)

31. Zettlemoyer and Collins (2005)

32. Zettlemoyer and Collins (2005)

33. NLP