~jtv/corpusfiltergraph/cross-python

« back to all changes in this revision

Viewing changes to trunk/lib/corpusfg/graphs/sa-champollion/config.ini

  • Committer: tahoar
  • Date: 2012-05-02 15:46:23 UTC
  • Revision ID: svn-v4:bc069b21-dff4-4e29-a776-06a4e04bad4e::266
new layout. need to update code to use the new layout

Show diffs side-by-side

added added

removed removed

Lines of Context:
 
1
[USER]
 
2
source=zh_cn
 
3
target=en
 
4
superdomains=*
 
5
domains=*
 
6
subdomains=*
 
7
filespec=*.txt
 
8
 
 
9
[manager]
 
10
debug=True
 
11
markreadonly=False
 
12
 
 
13
[filtergraph]
 
14
superdomains=%(superdomains)s
 
15
domains=%(domains)s
 
16
subdomains=%(subdomains)s
 
17
corpustypes=tm
 
18
languagepairs=%(source)s-%(target)s
 
19
filespec=%(filespec)s
 
20
 
 
21
[0,reader-file]
 
22
roottype=CORPORA
 
23
stage=fa
 
24
 
 
25
[0,writer-file]
 
26
roottype=CORPORA
 
27
stage=sa
 
28
 
 
29
[zh_cn]
 
30
filter0=0,remove-irrelevant
 
31
filter1=0,special-split
 
32
filter2=3,utf82mixed
 
33
filter3=3,utf8-num2arabic
 
34
filter4=3,mansegment-utf8
 
35
filter5=3,cn-tk-number-ascii
 
36
filter6=0,remove-multi-blanks
 
37
filter7=0,remove-blanks-start-end
 
38
;aligner0=2,aligner-champollion
 
39
 
 
40
[en]
 
41
filter0=0,remove-irrelevant
 
42
filter1=0,special-split
 
43
filter2=3,etoken
 
44
filter3=3,english_stemmer
 
45
filter4=0,convert-lowercase
 
46
filter5=0,remove-multi-blanks
 
47
filter6=0,remove-blanks-start-end
 
48
aligner0=2,aligner-champollion
 
49
 
 
50
;[ar]
 
51
;filter0=0,remove-irrelevant
 
52
;filter1=0,special-split
 
53
;filter2=3,arabic2mixed
 
54
;filter3=3,atoken
 
55
;filter4=3,ar-stemmer-utf8
 
56
;filter5=0,convert-lowercase
 
57
;filter6=0,remove-multi-blanks
 
58
;filter7=0,remove-blanks-start-end
 
59
;;aligner0=2,aligner-champollion
 
60
 
 
61
;[hn]
 
62
;filter0=0,remove-irrelevant
 
63
;filter1=0,special-split
 
64
;filter2=3,Htokenize
 
65
;filter3=3,light_hindi_stemmer
 
66
;filter4=3,romanize_hin_num
 
67
;filter5=0,convert-lowercase
 
68
;filter6=0,remove-multi-blanks
 
69
;filter7=0,remove-blanks-start-end
 
70
;;aligner0=2,aligner-champollion
 
71
 
 
72
[en,0,2,aligner-champollion]
 
73
invertX2Y=True
 
74
x2y=0.634
 
75
;       -c : number of Y chars for each X char (system autmatically inverts with invertX2Y=True)
 
76
simple-only=False
 
77
;       -n : disallow 1-3, 3-1, 1-4, 4-1 alignments
 
78
fast=False
 
79
;       -f : faster performance, lower accuracy
 
80
no-omissions=False
 
81
;       -a : align all and merge ommision
 
82
lowmem=False
 
83
;       -l : use on-disk database to reduce memory usage
 
84
 
 
85
; Do Moses Yourself Community Edition
 
86
; CorpusFiltergraph™ v4.0.264
 
87
; Copyright © 2010-2012 Precision Translation Tools Co., Ltd.
 
88
;      http://www.precisiontranslationtools.com
 
89
;      support@precisiontranslationtools.com