Modelling Natural Language with Finite Automata

1
Transducers Hinrich Schütze IMS, Uni Stuttgart,
WS 2007/08 Most slides borrowed from K. Haenelt
and E. Gurari
2
(No Transcript)
3
(No Transcript)
4
Diagram?
5
(No Transcript)
6

• Compute output for aabba

7
(No Transcript)
8

• Numbers to Numerals for English (1-99)

9
A morphological example
l
e
a
v
e
e
leave
l
e
a
v
s
e
leave
leaves
e
e
left
left
f
t
10
Stemanalysis input
noun-stem book ltnoun-stemgt ltnoun-suffixgt work ltno
un-stemgt ltnoun-suffixgt
e
book
e
e
N
work
e
book
eV
e
verb-stem book ltverb-suffixgt work ltverb-suffixgt
V
work
noun-suffix (eN Sg) (sN Pl)
verb-suffix (eV) (edV past) (sV
3rd)
11
A complex example

• noun-stem
• dog ltnoun-suffixgt
• fox ltnoun-suffixgt

noun-suffix (eN Sg) (sN Pl)
Transducer?
12
Modelling of WordsPhon./Orth. Alternation Rules
e-insertion rule for English plural nouns ending
with x,s,z (foxes)
,other
r5

other
z,s,x
s
z,s,x
z,s,x
s
r0
r1
r2
r3
r4
z,x
,other

,other
Jurafsky/Martin, 2000, S. 78
13
Modelling of WordsPhon./Orth. Alternation Rules
dictionary
oo
xx
ff
1
2
Ne
Pl
es
e
0
5
6
7
8
9
dd
oo
gg
3
4
rule
,other
5
e
z,s,x
other
,
e
z,s,x
s
D
W
ee
e
s
z,s,x
1
2
3
4
0
,other
z,x

,other
dictionary ? rule


xx
e
ee
ss

61
51
20
10
72
73
84

f
f
oo
xx
Ne
Pl
es
e
00
90


e
ss


30
81
40
50
60
70
dd
Pl
oo
gg
Ne
es
e
14

• DMOR/SMOR