shalmaneser 1.2.0.rc1 → 1.2.0.rc2
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/README.md +26 -8
- data/doc/SB_README +57 -0
- data/doc/exp_files_description.txt +160 -0
- data/doc/fred.pdf +0 -0
- data/doc/index.md +120 -0
- data/doc/salsa_tool.pdf +0 -0
- data/doc/salsatigerxml.pdf +0 -0
- data/doc/shal_doc.pdf +0 -0
- data/doc/shal_lrec.pdf +0 -0
- data/lib/ext/maxent/Classify.class +0 -0
- data/lib/ext/maxent/Train.class +0 -0
- data/lib/frprep/TreetaggerInterface.rb +4 -4
- data/lib/shalmaneser/version.rb +1 -1
- metadata +41 -48
- data/test/frprep/test_opt_parser.rb +0 -94
- data/test/functional/functional_test_helper.rb +0 -40
- data/test/functional/sample_experiment_files/fred_test.salsa.erb +0 -122
- data/test/functional/sample_experiment_files/fred_train.salsa.erb +0 -135
- data/test/functional/sample_experiment_files/prp_test.salsa.erb +0 -138
- data/test/functional/sample_experiment_files/prp_test.salsa.fred.standalone.erb +0 -120
- data/test/functional/sample_experiment_files/prp_test.salsa.rosy.standalone.erb +0 -120
- data/test/functional/sample_experiment_files/prp_train.salsa.erb +0 -138
- data/test/functional/sample_experiment_files/prp_train.salsa.fred.standalone.erb +0 -138
- data/test/functional/sample_experiment_files/prp_train.salsa.rosy.standalone.erb +0 -138
- data/test/functional/sample_experiment_files/rosy_test.salsa.erb +0 -257
- data/test/functional/sample_experiment_files/rosy_train.salsa.erb +0 -259
- data/test/functional/test_fred.rb +0 -47
- data/test/functional/test_frprep.rb +0 -52
- data/test/functional/test_rosy.rb +0 -40
@@ -1,135 +0,0 @@
|
|
1
|
-
# ID identifying this experiment and all its data
|
2
|
-
# please do not use spaces inside the experiment ID
|
3
|
-
experiment_ID = fred_train
|
4
|
-
|
5
|
-
# targets:
|
6
|
-
# if apply_to_all_known_targets is set to true,
|
7
|
-
# disambiguate all words for which we have training data
|
8
|
-
# when performing task "test" (i.e. applying trained classifiers)
|
9
|
-
apply_to_all_known_targets = true
|
10
|
-
|
11
|
-
# Enduser mode?
|
12
|
-
# The idea is that the enduser will only _apply_
|
13
|
-
# pre-trained classifiers. So in enduser mode many
|
14
|
-
# options are disallowed.
|
15
|
-
enduser_mode = false
|
16
|
-
|
17
|
-
|
18
|
-
# print warnings and
|
19
|
-
# give detailed progress reports
|
20
|
-
verbose = true
|
21
|
-
|
22
|
-
|
23
|
-
############################
|
24
|
-
# Paths
|
25
|
-
# - fred_directory: directory where Fred puts its internal data
|
26
|
-
# - directory_output:
|
27
|
-
# redirect system output of disambiguated text (in SalsaTigerXML)
|
28
|
-
# to another directory.
|
29
|
-
# If you do not set anything here, output is to
|
30
|
-
# <fred_directory>/<experiment_ID>/output/stxml
|
31
|
-
# - classifier_dir:
|
32
|
-
# Write trained classifiers to this directory.
|
33
|
-
# If you do not set this parameter, classifiers are written to
|
34
|
-
# <fred_directory>/<experiment_ID>/classifiers
|
35
|
-
|
36
|
-
fred_directory = <%= File.expand_path('test/functional/output') %>
|
37
|
-
|
38
|
-
# - preproc_descr_file_train / ...test
|
39
|
-
# where the experiment file for frprep is located
|
40
|
-
# (preprocessing for Fred and Rosy)
|
41
|
-
# for the preprocessing of the data used in this experiment
|
42
|
-
#
|
43
|
-
# give one preprocessing file name for the training data
|
44
|
-
# and one for the test data
|
45
|
-
# (If you only ever use test data in this experiment, you only
|
46
|
-
# need to give preproc_descr_file_test, and vice versa for training data.)
|
47
|
-
preproc_descr_file_train = <%= File.expand_path('test/functional/sample_experiment_files/prp_train.salsa.fred.standalone') %>
|
48
|
-
|
49
|
-
|
50
|
-
|
51
|
-
#####################
|
52
|
-
# noncontiguous input?
|
53
|
-
# if so, set 'noncontiguous_input' to 'true' (default is 'false')
|
54
|
-
# Also give the larger corpus from which the input sentences are:
|
55
|
-
# - directory
|
56
|
-
# - format: same possibilities as for frprep format
|
57
|
-
# - encoding: same possibilities as for frprep encoding
|
58
|
-
|
59
|
-
noncontiguous_input = false
|
60
|
-
#larger_corpus_dir =
|
61
|
-
larger_corpus_format = SalsaTigerXML
|
62
|
-
#larger_corpus_encoding = iso
|
63
|
-
|
64
|
-
|
65
|
-
#################
|
66
|
-
# Features
|
67
|
-
|
68
|
-
# bag-of-words context, with given context size,
|
69
|
-
# for example:
|
70
|
-
feature = context 50
|
71
|
-
feature = context 2
|
72
|
-
#
|
73
|
-
# (you can give more than one context feature line!)
|
74
|
-
#
|
75
|
-
# other possible features:
|
76
|
-
# feature = syntax
|
77
|
-
# feature = synsem
|
78
|
-
#
|
79
|
-
# syntax: grammatical functions
|
80
|
-
# synsem: grammatical functions plus headwords
|
81
|
-
|
82
|
-
#feature = context % %contextsize%
|
83
|
-
feature = syntax
|
84
|
-
|
85
|
-
# How to handle training data that is labeled
|
86
|
-
# with multiple sense labels?
|
87
|
-
# - binarize (default): This works only with binary classifiers.
|
88
|
-
# When featurizing for the binary classifiers, consider an item
|
89
|
-
# positive if its set of assigned labels includes the
|
90
|
-
# label for this binary classifier.
|
91
|
-
# - repeat: Repeat the instance, once for each
|
92
|
-
# sense label that has been assigned. (Basically, treat it
|
93
|
-
# as N instances with equal features but different labels.)
|
94
|
-
# - join: join all the assigned senses into one combined sense
|
95
|
-
# and treat that as a separate sense to train on.
|
96
|
-
# - keep: keep as multiple sense labels. (Note that this
|
97
|
-
# makes sense only for classifiers that can deal with
|
98
|
-
# multiple labels.)
|
99
|
-
|
100
|
-
#handle_multilabel = binarize
|
101
|
-
handle_multilabel = repeat
|
102
|
-
|
103
|
-
# What to do with numerical features?
|
104
|
-
# - keep: just leave as is
|
105
|
-
# - repeat: for a feature with max. numerical value N,
|
106
|
-
# use N binary features
|
107
|
-
# - bin: use a fixed number of bins, e.g. 5, then
|
108
|
-
# if feature value > 20: set all bins to 1,
|
109
|
-
# if feature value > 10: set the first four bins to 1,
|
110
|
-
# etc.
|
111
|
-
# default: bin.
|
112
|
-
#numerical_features = bin
|
113
|
-
numerical_features = keep
|
114
|
-
# Binary classifiers, or n-ary classifiers?
|
115
|
-
# if binary classifiers, set 'binary_classifiers = true'
|
116
|
-
# default is 'false'.
|
117
|
-
binary_classifiers = false
|
118
|
-
|
119
|
-
#################
|
120
|
-
# Fred internal settings
|
121
|
-
|
122
|
-
# what kind of classifier to use?
|
123
|
-
#
|
124
|
-
# format:
|
125
|
-
# <classifier type> <path> <optionally another path>
|
126
|
-
#
|
127
|
-
# for maxent, give first the path where maxent resides,
|
128
|
-
# then <where_shalmaneser_resides>/program/tools/maxent
|
129
|
-
classifier = maxent <%= File.expand_path('tools/maxent/maxent-2.4.0') %>
|
130
|
-
|
131
|
-
|
132
|
-
# for binary classifiers, you can set the pseudolabel
|
133
|
-
# on the 'negative' sense.
|
134
|
-
# Default is 'NONE'
|
135
|
-
negsense = NONE
|
@@ -1,138 +0,0 @@
|
|
1
|
-
#################################################
|
2
|
-
# This is a sample experiment file
|
3
|
-
# with explanations of all features
|
4
|
-
# that can be set for the frprep preprocessing system for Fred and Rosy.
|
5
|
-
#
|
6
|
-
# To start your own experiment,
|
7
|
-
# replace all occurrences of
|
8
|
-
# %...% by values of your choice.
|
9
|
-
#
|
10
|
-
# Boolean features may be omitted and are false by default.
|
11
|
-
#
|
12
|
-
# Experiment file lines that start with '#'
|
13
|
-
# are comments and are ignored. Empty lines are ignored as well.
|
14
|
-
|
15
|
-
########################
|
16
|
-
# Experiment description
|
17
|
-
#
|
18
|
-
|
19
|
-
# ID identifying this experiment and all its data
|
20
|
-
# please do not use spaces inside the experiment ID
|
21
|
-
prep_experiment_ID = prp_test
|
22
|
-
|
23
|
-
# YOUR INPUT DATA:
|
24
|
-
# frprep accepts an input directory rather than an input file.
|
25
|
-
# It will process all files in the directory directory_input
|
26
|
-
# and write the results to directory_preprocessed.
|
27
|
-
#
|
28
|
-
# For input formats see the discussion of "format" below.
|
29
|
-
directory_input = <%= File.expand_path('test/functional/input/frprep/test.salsa') %>
|
30
|
-
directory_preprocessed = <%= File.expand_path('test/functional/output/frprep/test.salsa') %>
|
31
|
-
|
32
|
-
##
|
33
|
-
# Experimental data is described by the following parameters:
|
34
|
-
#
|
35
|
-
# - language: en / de
|
36
|
-
# en for English or de for German
|
37
|
-
#
|
38
|
-
# - format: SalsaTigerXML / FNXml / SalsaTab / BNC / Plain
|
39
|
-
#
|
40
|
-
# Format of the input data, training/test set
|
41
|
-
# SalsaTigerXML: Parsed data, English or German
|
42
|
-
# FNXml: FrameNet Lexical Unit files in FrameNet XML format
|
43
|
-
# FNCorpusXML: FrameNet files in the FrameNet corpus XML format
|
44
|
-
# SalsaTab: tabular format (internal)
|
45
|
-
# BNC BNC XML format, alternating words and POS tags
|
46
|
-
# Plain Plain text, ONE SENTENCE PER LINE.
|
47
|
-
#
|
48
|
-
# Preprocessing transforms all data to SalsaTigerXML.
|
49
|
-
#
|
50
|
-
# - origin: SalsaTiger / FrameNet / <not specified>
|
51
|
-
# This is the origin of the training/test data.
|
52
|
-
# SalsaTiger: data from the Tiger corpus, possibly semantically
|
53
|
-
# annotated by Salsa
|
54
|
-
# FrameNet: data from the FrameNet project
|
55
|
-
#
|
56
|
-
# Don't set 'origin' if none of these origins apply
|
57
|
-
#
|
58
|
-
# - encoding: utf8 / iso / hex / <not specified>
|
59
|
-
# Default: iso
|
60
|
-
|
61
|
-
language = de
|
62
|
-
#origin =
|
63
|
-
format = Plain
|
64
|
-
encoding = iso
|
65
|
-
|
66
|
-
#############################
|
67
|
-
# Which preprocessing steps to take?
|
68
|
-
#
|
69
|
-
# Data can be parsed, lemmatized and POS-tagged,
|
70
|
-
# but this happens only if it is specified in the
|
71
|
-
# experiment file.
|
72
|
-
#
|
73
|
-
# Set these booleans to true to trigger the respective
|
74
|
-
# type of preprocessing. The default value is false.
|
75
|
-
|
76
|
-
do_lemmatize = true
|
77
|
-
do_postag = false
|
78
|
-
do_parse = true
|
79
|
-
|
80
|
-
#############################
|
81
|
-
# directory where frprep puts its internal data
|
82
|
-
#
|
83
|
-
|
84
|
-
frprep_directory = <%= File.expand_path('test/functional/output/') %>
|
85
|
-
|
86
|
-
#############################
|
87
|
-
# Syntax/semantics interface repair:
|
88
|
-
# FrameNet annotated data has some annotation choices
|
89
|
-
# that may make it harder to learn the mapping from
|
90
|
-
# syntactic structure to semantic roles.
|
91
|
-
#
|
92
|
-
# If you are using FrameNet data for training a
|
93
|
-
# semantic role labeler, set the following two settings
|
94
|
-
# to true (default is false) to 'repair' semantic role labels
|
95
|
-
# to closer match the syntactic structure
|
96
|
-
|
97
|
-
fe_syn_repair = true
|
98
|
-
fe_rel_repair = false
|
99
|
-
|
100
|
-
|
101
|
-
#################
|
102
|
-
# Location of tools and resources used by Fred
|
103
|
-
|
104
|
-
# currently known to the system:
|
105
|
-
# (Saarbruecken paths given)
|
106
|
-
#
|
107
|
-
# - POS tagging:
|
108
|
-
# - pos_tagger = treetagger
|
109
|
-
# pos_tagger_path = /proj/llx/Software/treetagger/cmd/tree-tagger-english-notokenisation
|
110
|
-
#
|
111
|
-
# - Lemmatization:
|
112
|
-
# - lemmatizer = treetagger
|
113
|
-
# lemmatizer_path = /proj/llx/Software/treetagger/cmd/tree-tagger-english-notokenisation
|
114
|
-
# lemmatizer_path = /proj/llx/Software/treetagger/cmd/tree-tagger-german-notokenisation
|
115
|
-
#
|
116
|
-
# - Parser:
|
117
|
-
# - parser = collins (English)
|
118
|
-
# parser_path = /proj/llx/Software/Parsers/COLLINS-PARSER/
|
119
|
-
# - parser = sleepy (German)
|
120
|
-
# parser_path = /proj/corpora/sleepy3/
|
121
|
-
# - parser = minipar (English)
|
122
|
-
# parser_path = /proj/llx/Software/Parsers/minipar-linux/
|
123
|
-
#
|
124
|
-
pos_tagger = treetagger
|
125
|
-
pos_tagger_path = <%= File.expand_path('tools/treetagger') %>
|
126
|
-
|
127
|
-
lemmatizer = treetagger
|
128
|
-
lemmatizer_path = <%= File.expand_path('tools/treetagger') %>
|
129
|
-
|
130
|
-
parser = berkeley
|
131
|
-
parser_path = <%= File.expand_path('tools/berkeleyParser') %>
|
132
|
-
|
133
|
-
# parser:
|
134
|
-
# maximum no. of sentences in a parse file,
|
135
|
-
# maximum sentence length to be parsed
|
136
|
-
|
137
|
-
parser_max_sent_num = 2000
|
138
|
-
parser_max_sent_len = 80
|
@@ -1,120 +0,0 @@
|
|
1
|
-
# ID identifying this experiment and all its data
|
2
|
-
# please do not use spaces inside the experiment ID
|
3
|
-
prep_experiment_ID = prp_test
|
4
|
-
|
5
|
-
# YOUR INPUT DATA:
|
6
|
-
# frprep accepts an input directory rather than an input file.
|
7
|
-
# It will process all files in the directory directory_input
|
8
|
-
# and write the results to directory_preprocessed.
|
9
|
-
#
|
10
|
-
# For input formats see the discussion of "format" below.
|
11
|
-
#directory_input = <%= File.expand_path('test/functional/input/frprep/test.salsa') %>
|
12
|
-
directory_preprocessed = <%= File.expand_path('test/functional/input/fred/frprep/test.salsa') %>
|
13
|
-
|
14
|
-
##
|
15
|
-
# Experimental data is described by the following parameters:
|
16
|
-
#
|
17
|
-
# - language: en / de
|
18
|
-
# en for English or de for German
|
19
|
-
#
|
20
|
-
# - format: SalsaTigerXML / FNXml / SalsaTab / BNC / Plain
|
21
|
-
#
|
22
|
-
# Format of the input data, training/test set
|
23
|
-
# SalsaTigerXML: Parsed data, English or German
|
24
|
-
# FNXml: FrameNet Lexical Unit files in FrameNet XML format
|
25
|
-
# FNCorpusXML: FrameNet files in the FrameNet corpus XML format
|
26
|
-
# SalsaTab: tabular format (internal)
|
27
|
-
# BNC BNC XML format, alternating words and POS tags
|
28
|
-
# Plain Plain text, ONE SENTENCE PER LINE.
|
29
|
-
#
|
30
|
-
# Preprocessing transforms all data to SalsaTigerXML.
|
31
|
-
#
|
32
|
-
# - origin: SalsaTiger / FrameNet / <not specified>
|
33
|
-
# This is the origin of the training/test data.
|
34
|
-
# SalsaTiger: data from the Tiger corpus, possibly semantically
|
35
|
-
# annotated by Salsa
|
36
|
-
# FrameNet: data from the FrameNet project
|
37
|
-
#
|
38
|
-
# Don't set 'origin' if none of these origins apply
|
39
|
-
#
|
40
|
-
# - encoding: utf8 / iso / hex / <not specified>
|
41
|
-
# Default: iso
|
42
|
-
|
43
|
-
language = de
|
44
|
-
#origin =
|
45
|
-
format = Plain
|
46
|
-
encoding = iso
|
47
|
-
|
48
|
-
#############################
|
49
|
-
# Which preprocessing steps to take?
|
50
|
-
#
|
51
|
-
# Data can be parsed, lemmatized and POS-tagged,
|
52
|
-
# but this happens only if it is specified in the
|
53
|
-
# experiment file.
|
54
|
-
#
|
55
|
-
# Set these booleans to true to trigger the respective
|
56
|
-
# type of preprocessing. The default value is false.
|
57
|
-
|
58
|
-
do_lemmatize = true
|
59
|
-
do_postag = false
|
60
|
-
do_parse = true
|
61
|
-
|
62
|
-
#############################
|
63
|
-
# directory where frprep puts its internal data
|
64
|
-
#
|
65
|
-
|
66
|
-
#frprep_directory = <%= File.expand_path('test/functional/input/fred/frprep') %>
|
67
|
-
|
68
|
-
#############################
|
69
|
-
# Syntax/semantics interface repair:
|
70
|
-
# FrameNet annotated data has some annotation choices
|
71
|
-
# that may make it harder to learn the mapping from
|
72
|
-
# syntactic structure to semantic roles.
|
73
|
-
#
|
74
|
-
# If you are using FrameNet data for training a
|
75
|
-
# semantic role labeler, set the following two settings
|
76
|
-
# to true (default is false) to 'repair' semantic role labels
|
77
|
-
# to closer match the syntactic structure
|
78
|
-
|
79
|
-
fe_syn_repair = true
|
80
|
-
fe_rel_repair = false
|
81
|
-
|
82
|
-
|
83
|
-
#################
|
84
|
-
# Location of tools and resources used by Fred
|
85
|
-
|
86
|
-
# currently known to the system:
|
87
|
-
# (Saarbruecken paths given)
|
88
|
-
#
|
89
|
-
# - POS tagging:
|
90
|
-
# - pos_tagger = treetagger
|
91
|
-
# pos_tagger_path = /proj/llx/Software/treetagger/cmd/tree-tagger-english-notokenisation
|
92
|
-
#
|
93
|
-
# - Lemmatization:
|
94
|
-
# - lemmatizer = treetagger
|
95
|
-
# lemmatizer_path = /proj/llx/Software/treetagger/cmd/tree-tagger-english-notokenisation
|
96
|
-
# lemmatizer_path = /proj/llx/Software/treetagger/cmd/tree-tagger-german-notokenisation
|
97
|
-
#
|
98
|
-
# - Parser:
|
99
|
-
# - parser = collins (English)
|
100
|
-
# parser_path = /proj/llx/Software/Parsers/COLLINS-PARSER/
|
101
|
-
# - parser = sleepy (German)
|
102
|
-
# parser_path = /proj/corpora/sleepy3/
|
103
|
-
# - parser = minipar (English)
|
104
|
-
# parser_path = /proj/llx/Software/Parsers/minipar-linux/
|
105
|
-
#
|
106
|
-
pos_tagger = treetagger
|
107
|
-
pos_tagger_path = <%= File.expand_path('tools/treetagger') %>
|
108
|
-
|
109
|
-
lemmatizer = treetagger
|
110
|
-
lemmatizer_path = <%= File.expand_path('tools/treetagger') %>
|
111
|
-
|
112
|
-
parser = berkeley
|
113
|
-
parser_path = <%= File.expand_path('tools/berkeleyParser') %>
|
114
|
-
|
115
|
-
# parser:
|
116
|
-
# maximum no. of sentences in a parse file,
|
117
|
-
# maximum sentence length to be parsed
|
118
|
-
|
119
|
-
parser_max_sent_num = 2000
|
120
|
-
parser_max_sent_len = 80
|
@@ -1,120 +0,0 @@
|
|
1
|
-
# ID identifying this experiment and all its data
|
2
|
-
# please do not use spaces inside the experiment ID
|
3
|
-
prep_experiment_ID = prp_test
|
4
|
-
|
5
|
-
# YOUR INPUT DATA:
|
6
|
-
# frprep accepts an input directory rather than an input file.
|
7
|
-
# It will process all files in the directory directory_input
|
8
|
-
# and write the results to directory_preprocessed.
|
9
|
-
#
|
10
|
-
# For input formats see the discussion of "format" below.
|
11
|
-
#directory_input = <%= File.expand_path('test/functional/input/frprep/test.salsa') %>
|
12
|
-
directory_preprocessed = <%= File.expand_path('test/functional/input/rosy/frprep/test.salsa') %>
|
13
|
-
|
14
|
-
##
|
15
|
-
# Experimental data is described by the following parameters:
|
16
|
-
#
|
17
|
-
# - language: en / de
|
18
|
-
# en for English or de for German
|
19
|
-
#
|
20
|
-
# - format: SalsaTigerXML / FNXml / SalsaTab / BNC / Plain
|
21
|
-
#
|
22
|
-
# Format of the input data, training/test set
|
23
|
-
# SalsaTigerXML: Parsed data, English or German
|
24
|
-
# FNXml: FrameNet Lexical Unit files in FrameNet XML format
|
25
|
-
# FNCorpusXML: FrameNet files in the FrameNet corpus XML format
|
26
|
-
# SalsaTab: tabular format (internal)
|
27
|
-
# BNC BNC XML format, alternating words and POS tags
|
28
|
-
# Plain Plain text, ONE SENTENCE PER LINE.
|
29
|
-
#
|
30
|
-
# Preprocessing transforms all data to SalsaTigerXML.
|
31
|
-
#
|
32
|
-
# - origin: SalsaTiger / FrameNet / <not specified>
|
33
|
-
# This is the origin of the training/test data.
|
34
|
-
# SalsaTiger: data from the Tiger corpus, possibly semantically
|
35
|
-
# annotated by Salsa
|
36
|
-
# FrameNet: data from the FrameNet project
|
37
|
-
#
|
38
|
-
# Don't set 'origin' if none of these origins apply
|
39
|
-
#
|
40
|
-
# - encoding: utf8 / iso / hex / <not specified>
|
41
|
-
# Default: iso
|
42
|
-
|
43
|
-
language = de
|
44
|
-
#origin =
|
45
|
-
format = Plain
|
46
|
-
encoding = iso
|
47
|
-
|
48
|
-
#############################
|
49
|
-
# Which preprocessing steps to take?
|
50
|
-
#
|
51
|
-
# Data can be parsed, lemmatized and POS-tagged,
|
52
|
-
# but this happens only if it is specified in the
|
53
|
-
# experiment file.
|
54
|
-
#
|
55
|
-
# Set these booleans to true to trigger the respective
|
56
|
-
# type of preprocessing. The default value is false.
|
57
|
-
|
58
|
-
do_lemmatize = true
|
59
|
-
do_postag = false
|
60
|
-
do_parse = true
|
61
|
-
|
62
|
-
#############################
|
63
|
-
# directory where frprep puts its internal data
|
64
|
-
#
|
65
|
-
|
66
|
-
#frprep_directory = <%= File.expand_path('test/functional/input/rosy/frprep') %>
|
67
|
-
|
68
|
-
#############################
|
69
|
-
# Syntax/semantics interface repair:
|
70
|
-
# FrameNet annotated data has some annotation choices
|
71
|
-
# that may make it harder to learn the mapping from
|
72
|
-
# syntactic structure to semantic roles.
|
73
|
-
#
|
74
|
-
# If you are using FrameNet data for training a
|
75
|
-
# semantic role labeler, set the following two settings
|
76
|
-
# to true (default is false) to 'repair' semantic role labels
|
77
|
-
# to closer match the syntactic structure
|
78
|
-
|
79
|
-
fe_syn_repair = true
|
80
|
-
fe_rel_repair = false
|
81
|
-
|
82
|
-
|
83
|
-
#################
|
84
|
-
# Location of tools and resources used by Fred
|
85
|
-
|
86
|
-
# currently known to the system:
|
87
|
-
# (Saarbruecken paths given)
|
88
|
-
#
|
89
|
-
# - POS tagging:
|
90
|
-
# - pos_tagger = treetagger
|
91
|
-
# pos_tagger_path = /proj/llx/Software/treetagger/cmd/tree-tagger-english-notokenisation
|
92
|
-
#
|
93
|
-
# - Lemmatization:
|
94
|
-
# - lemmatizer = treetagger
|
95
|
-
# lemmatizer_path = /proj/llx/Software/treetagger/cmd/tree-tagger-english-notokenisation
|
96
|
-
# lemmatizer_path = /proj/llx/Software/treetagger/cmd/tree-tagger-german-notokenisation
|
97
|
-
#
|
98
|
-
# - Parser:
|
99
|
-
# - parser = collins (English)
|
100
|
-
# parser_path = /proj/llx/Software/Parsers/COLLINS-PARSER/
|
101
|
-
# - parser = sleepy (German)
|
102
|
-
# parser_path = /proj/corpora/sleepy3/
|
103
|
-
# - parser = minipar (English)
|
104
|
-
# parser_path = /proj/llx/Software/Parsers/minipar-linux/
|
105
|
-
#
|
106
|
-
pos_tagger = treetagger
|
107
|
-
pos_tagger_path = <%= File.expand_path('tools/treetagger') %>
|
108
|
-
|
109
|
-
lemmatizer = treetagger
|
110
|
-
lemmatizer_path = <%= File.expand_path('tools/treetagger') %>
|
111
|
-
|
112
|
-
parser = berkeley
|
113
|
-
parser_path = <%= File.expand_path('tools/berkeleyParser') %>
|
114
|
-
|
115
|
-
# parser:
|
116
|
-
# maximum no. of sentences in a parse file,
|
117
|
-
# maximum sentence length to be parsed
|
118
|
-
|
119
|
-
parser_max_sent_num = 2000
|
120
|
-
parser_max_sent_len = 80
|