Proof-nets and semantic applications - PowerPoint PPT Presentation

About This Presentation
Title:

Proof-nets and semantic applications

Description:

Proof-nets and semantic applications. Alain Lecomte. ESSLLI2002 ... plugging lexical semantic types to the homomorphic PN by cut (!e o t) o ((!e o t) ot) ... – PowerPoint PPT presentation

Number of Views:21
Avg rating:3.0/5.0
Slides: 69
Provided by: AlainL66
Category:

less

Transcript and Presenter's Notes

Title: Proof-nets and semantic applications


1
Proof-nets and semantic applications
  • Alain Lecomte
  • ESSLLI2002

2
Semantic proof nets
  • child

xe, child e?t - child(x) t hence child
e?t - ?x.child(x)e?t
3
  • run

e
t-
x
e-
t
?x
child
4
  • run

e
t-
x
e-
t
?x
child
5
  • run

e
t-
x
e-
t
?x
child
6
  • run

e
t-
x
e-
t
x
?x
child
?x.child(x)
7
  • run

e
t-
x
e-
t
x
?x
child
?x.child(x)
8
each, every
  • A determiner like every, each decomposes into
  • A quantifier, for instance ?
  • type (e?t)?t
  • A connective, for instance ?
  • type t?(t?t)

9
  • needs two predicates (e ? t) for obtaining one
    proposition (t)
  • A determiner is therefore of type
  • (e?t)?((e?t)?t)

10
  • A determiner is therefore associated with a
    sequent
  • Its semantic is represented by its proof

11
deduction
12
remark
  • With a very remarkable step an application of
    the contraction rule!
  • ? necessity of working inside Intuitionistic
    linear logic with exponentials
  • The exact sequent which encodes the determiner is

!e
!e
)
)
((
)
(
)
(
),
(
t
t
t
t
t
!e
t
t
t
--
--o
--o
--o
--o
--o
--o
--o
--o
13
Exponentials
14
Exponentials (one-sided)
15
Representation of the proof
c
?
?
(!e o t) o ((!e o t) o t)
16
every child
c
child
(!e o t) o t
?
?
17
every child likes to play
c
likes to play
t
child
?
?
18
Application

-

A
A o B
19
Application
A -
B
20
Abstraction
B -
21
Abstraction
B -
B o A
22
Syntactic proof-nets
  • Proof-nets for Lambek calculus
  • Like PN for MILL
  • condition on semi-planarity

23
every child plays
  • 1) unfolding

np -
s
s -
np\s
np
(s/(np\s)) -
n
s -
n - child
np\s - plays
s
(s/(np\s))/n - every
24
every child plays
  • 2) links

np -
s
s -
np\s
np
(s/(np\s)) -
n
s -
n - child
np\s - plays
s
(s/(np\s))/n - every
25
Attention!
  • 2) links

WRONG !
np -
s
s -
np\s
np
(s/(np\s)) -
n
s -
n - child
np\s - plays
s
(s/(np\s))/n - every
26
Parsing
  • through homomorphism
  • H(s) t
  • H(np) !e
  • H(n) !e o t
  • H(A/B) H(B\A) H(B) o H(A)

27
every child plays

np -
s
s -
np\s
np
(s/(np\s)) -
n
s -
n - child
np\s - plays
s
(s/(np\s))/n - every
28
every child plays
  • 3) homomorphism

!e
t
t
!e o t
!e
(!e o t) o t
!e o t
t
!e o t child
!e o t plays
t
(!e o t) o ((!e o t) ot)) every
29
  • semantic recipes
  • child ?x.child(x)
  • every ?P.?Q.?(?x.(P(x)?Q(x))
  • plays ?x.play(x)

30
  • represented by proof-nets

31
  • represented by proof-nets

d
32
every
c
?
?
(!e o t) o ((!e o t) o t)
33
plugging lexical semantic types to the
homomorphic PN by cut
34

!e
t
t
!e o t
!e
(!e o t) o t
!e o t
t
!e o t child
!e o t plays
t
(!e o t) o ((!e o t) ot)) every
CUT
35

!e
t
t
!e o t
!e
t
!e
(!e o t) o t
t
!e o t plays
t
(!e o t) o ((!e o t) ot)) every
d
child
36

!e
t
t
!e o t
!e
t
!e
(!e o t) o t
t
!e o t plays
t
(!e o t) o ((!e o t) ot)) every
d
child
CUT
37

!e
t
d
t
!e o t
!e
t
!e
(!e o t) o t
t
plays
t
(!e o t) o ((!e o t) ot)) every
d
child
38

PNevery
!e
t
d
t
!e o t
!e
t
!e
(!e o t) o t
t
plays
t
(!e o t) o ((!e o t) ot)) every
d
child
CUT
39

!e
t
c
plays
t
d
?
?
child
40

!e
t
c
plays
t
d
?
?
child
41

!e
t
c
plays
t
d
?
?
?
child
42

!e
t
c
plays
?x
t
d
?
?
?
child
43

!e
t
c
plays
?x
t
d
?
?
?
?
child
44

!e
t
c
plays
?x
plays
t
d
?
?
?
?
child
45

!e
t
c
plays
?x
plays
t
d
?
?
child
?
?
child
46

child(x)
!e
t
c
plays
?x
plays
t
d
?
?
child
?
?
child
47

plays(x)
child(x)
!e
t
c
plays
?x
plays
t
d
?
?
child
?(?x.(?(child(x),plays(x))))
?
?
child
48
Logical synthesisfrom a formula to a sentence
  • the reverse story
  • Start
  • a semantic formula ?
  • semantic recipes for lexical entries ?1, ?2,
    ?n
  • Goal
  • A sentence using all these recipes the meaning of
    which is ?

49
Usual solutions?-term unification
? skiss(p,m)
Peter np p
kisses (np\s)/np ?x.?y.kiss(y,x)
Mary np m
GOAL
50
s-
np
?
kiss(?,?)
np
?y.kiss(y, ?)
?
? skiss(p,m)
Peter np- p
kisses (np\s)/np ?x.?y.kiss(y,x)
Mary np- m
51
s-
np
kiss(?,?)
?
np
?y.kiss(y, ?)
?
? skiss(p,m)
Peter np- p
kisses (np\s)/np ?x.?y.kiss(y,x)
Mary np- m
52
kiss(?,?) kiss(p,m) ? ? m ? p
s-
np
?
kiss(?,?)
np
?y.kiss(y, ?)
?
? skiss(p,m)
Peter np- p
kisses (np\s)/np ?x.?y.kiss(y,x)
Mary np- m
53
kiss(?,?) kiss(p,m) ? ? m ? p
? m
s-
np
?
kiss(?,?)
np
?y.kiss(y, ?)
?
? skiss(p,m)
Peter np- p
kisses (np\s)/np ?x.?y.kiss(y,x)
Mary np- m
54
kiss(?,?) kiss(p,m) ? ? m ? p
? m
? p
s-
np
?
kiss(?,?)
np
?y.kiss(y, ?)
?
? skiss(p,m)
Peter np- p
kisses (np\s)/np ?x.?y.kiss(y,x)
Mary np- m
55
The net is a proof-net for L, therefore the
correct sentence is Peter kisses Mary
kiss(?,?) kiss(p,m) ? ? m ? p
? m
? p
s-
np
?
kiss(?,?)
np
?y.kiss(y, ?)
?
? skiss(p,m)
Peter np- p
kisses (np\s)/np ?x.?y.kiss(y,x)
Mary np- m
56
But
  • Non decidability for second order
  • xf f(fa) ?
  • no mgu
  • two incomparable solutions
  • ?z.z(fa)
  • ?z.f(za)

57
the use of PN
  • Generation (or synthesis )
  • research of a proof in L (which is decidable!),
    helped by a semantic form

58
The problem
  • Find PNL such that
  • plug by cuts lexical PNs PN?1, PN?2, PN?n
  • cut-elimination
  • Given PN?

59
?
APNL
T1
T2
Tn
60
?
AH(PNL)
HT1
HT2
HTn
CUT
CUT
61
PN?
62
the execution formula
  • Let P be a PN, U the set of its axiom links, ?
    the set of its cut links
  • u incidence matrix of U
  • ? incidence matrix of ?

63

Cut elimination between axioms
e1
e2
e3
e4
e1
e2
e3
e4
e1
e2
e3
e4
- X, X?
- X, X?
CUT
- X, X?
64
  • after a first step of cut-elimination on axioms
    u replaced by u?u
  • links coming from cut-elimination of level 1
  • To suppress all the links the premisses of which
    are premisses of a new cut and all the links
    which had no incident cut
  • u?u - ?2u?u? - u?u?2 ?2u?2
  • (1- ?2)u?u(1- ?2)

65
  • Idem for links coming from elimination of level
    2, 3, , n, cuts
  • Resulting graph after cut-elimination

66
?
U
HT1
HT2
HTn
CUT
?
CUT
67
Res(U,?)
68
  • U can be calculated from Res(U,?) and ?
  • Cf. PhD thesis by Sylvain POGODALLA
  • http//www.xrce.xerox.com/pogodalla
  • Condition that each lexical semantics contain
    at least one constant which intervenes in the
    global semantic representation.
Write a Comment
User Comments (0)
About PowerShow.com