opener-opinion-detector-base 2.0.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/README.md +101 -0
- data/bin/opinion-detector-base +19 -0
- data/core/annotation.cfg.erb +9 -0
- data/core/packages/KafNafParser-1.4.tar.gz +0 -0
- data/core/packages/VUA_pylib-1.5.tar.gz +0 -0
- data/core/python-scripts/LICENSE +339 -0
- data/core/python-scripts/README.md +226 -0
- data/core/python-scripts/classify_kaf_naf_file.py +499 -0
- data/core/python-scripts/cross_validation.py +634 -0
- data/core/python-scripts/generate_folds.py +134 -0
- data/core/python-scripts/models.cfg +10 -0
- data/core/python-scripts/my_templates/README +33 -0
- data/core/python-scripts/my_templates/templates_exp.only0.txt +6 -0
- data/core/python-scripts/my_templates/templates_exp.pol0.txt +10 -0
- data/core/python-scripts/my_templates/templates_exp.red.txt +7 -0
- data/core/python-scripts/my_templates/templates_exp.txt +10 -0
- data/core/python-scripts/my_templates/templates_holder.only0.txt +11 -0
- data/core/python-scripts/my_templates/templates_holder.red.txt +9 -0
- data/core/python-scripts/my_templates/templates_holder.txt +10 -0
- data/core/python-scripts/my_templates/templates_target.only0.txt +11 -0
- data/core/python-scripts/my_templates/templates_target.red.txt +9 -0
- data/core/python-scripts/my_templates/templates_target.txt +10 -0
- data/core/python-scripts/run_all_experiments.sh +49 -0
- data/core/python-scripts/run_basic.py +20 -0
- data/core/python-scripts/run_experiment.sh +42 -0
- data/core/python-scripts/scripts/__init__.py +1 -0
- data/core/python-scripts/scripts/config_manager.py +314 -0
- data/core/python-scripts/scripts/crfutils.py +215 -0
- data/core/python-scripts/scripts/extract_feats_relations.py +295 -0
- data/core/python-scripts/scripts/extract_features.py +376 -0
- data/core/python-scripts/scripts/feats_to_crf.exp.py +105 -0
- data/core/python-scripts/scripts/lexicons.py +44 -0
- data/core/python-scripts/scripts/link_entities_distance.py +77 -0
- data/core/python-scripts/scripts/relation_classifier.py +250 -0
- data/core/python-scripts/train.py +566 -0
- data/core/site-packages/pre_build/KafNafParser-1.4-py2.7.egg-info/PKG-INFO +10 -0
- data/core/site-packages/pre_build/KafNafParser-1.4-py2.7.egg-info/SOURCES.txt +22 -0
- data/core/site-packages/pre_build/KafNafParser-1.4-py2.7.egg-info/dependency_links.txt +1 -0
- data/core/site-packages/pre_build/KafNafParser-1.4-py2.7.egg-info/installed-files.txt +47 -0
- data/core/site-packages/pre_build/KafNafParser-1.4-py2.7.egg-info/top_level.txt +1 -0
- data/core/site-packages/pre_build/KafNafParser/KafNafParserMod.py +390 -0
- data/core/site-packages/pre_build/KafNafParser/KafNafParserMod.pyc +0 -0
- data/core/site-packages/pre_build/KafNafParser/__init__.py +14 -0
- data/core/site-packages/pre_build/KafNafParser/__init__.pyc +0 -0
- data/core/site-packages/pre_build/KafNafParser/constituency_data.py +125 -0
- data/core/site-packages/pre_build/KafNafParser/constituency_data.pyc +0 -0
- data/core/site-packages/pre_build/KafNafParser/coreference_data.py +52 -0
- data/core/site-packages/pre_build/KafNafParser/coreference_data.pyc +0 -0
- data/core/site-packages/pre_build/KafNafParser/dependency_data.py +78 -0
- data/core/site-packages/pre_build/KafNafParser/dependency_data.pyc +0 -0
- data/core/site-packages/pre_build/KafNafParser/entity_data.py +59 -0
- data/core/site-packages/pre_build/KafNafParser/entity_data.pyc +0 -0
- data/core/site-packages/pre_build/KafNafParser/external_references_data.py +41 -0
- data/core/site-packages/pre_build/KafNafParser/external_references_data.pyc +0 -0
- data/core/site-packages/pre_build/KafNafParser/feature_extractor/__init__.py +2 -0
- data/core/site-packages/pre_build/KafNafParser/feature_extractor/__init__.pyc +0 -0
- data/core/site-packages/pre_build/KafNafParser/feature_extractor/constituency.py +205 -0
- data/core/site-packages/pre_build/KafNafParser/feature_extractor/constituency.pyc +0 -0
- data/core/site-packages/pre_build/KafNafParser/feature_extractor/dependency.py +309 -0
- data/core/site-packages/pre_build/KafNafParser/feature_extractor/dependency.pyc +0 -0
- data/core/site-packages/pre_build/KafNafParser/features_data.py +131 -0
- data/core/site-packages/pre_build/KafNafParser/features_data.pyc +0 -0
- data/core/site-packages/pre_build/KafNafParser/header_data.py +127 -0
- data/core/site-packages/pre_build/KafNafParser/header_data.pyc +0 -0
- data/core/site-packages/pre_build/KafNafParser/opinion_data.py +211 -0
- data/core/site-packages/pre_build/KafNafParser/opinion_data.pyc +0 -0
- data/core/site-packages/pre_build/KafNafParser/references_data.py +23 -0
- data/core/site-packages/pre_build/KafNafParser/references_data.pyc +0 -0
- data/core/site-packages/pre_build/KafNafParser/span_data.py +63 -0
- data/core/site-packages/pre_build/KafNafParser/span_data.pyc +0 -0
- data/core/site-packages/pre_build/KafNafParser/term_data.py +111 -0
- data/core/site-packages/pre_build/KafNafParser/term_data.pyc +0 -0
- data/core/site-packages/pre_build/KafNafParser/term_sentiment_data.py +42 -0
- data/core/site-packages/pre_build/KafNafParser/term_sentiment_data.pyc +0 -0
- data/core/site-packages/pre_build/KafNafParser/text_data.py +99 -0
- data/core/site-packages/pre_build/KafNafParser/text_data.pyc +0 -0
- data/core/site-packages/pre_build/VUA_pylib-1.5-py2.7.egg-info/PKG-INFO +10 -0
- data/core/site-packages/pre_build/VUA_pylib-1.5-py2.7.egg-info/SOURCES.txt +14 -0
- data/core/site-packages/pre_build/VUA_pylib-1.5-py2.7.egg-info/dependency_links.txt +1 -0
- data/core/site-packages/pre_build/VUA_pylib-1.5-py2.7.egg-info/installed-files.txt +23 -0
- data/core/site-packages/pre_build/VUA_pylib-1.5-py2.7.egg-info/top_level.txt +1 -0
- data/core/site-packages/pre_build/VUA_pylib/__init__.py +1 -0
- data/core/site-packages/pre_build/VUA_pylib/__init__.pyc +0 -0
- data/core/site-packages/pre_build/VUA_pylib/common/__init__.py +1 -0
- data/core/site-packages/pre_build/VUA_pylib/common/__init__.pyc +0 -0
- data/core/site-packages/pre_build/VUA_pylib/common/common.py +28 -0
- data/core/site-packages/pre_build/VUA_pylib/common/common.pyc +0 -0
- data/core/site-packages/pre_build/VUA_pylib/corpus_reader/__init__.py +1 -0
- data/core/site-packages/pre_build/VUA_pylib/corpus_reader/__init__.pyc +0 -0
- data/core/site-packages/pre_build/VUA_pylib/corpus_reader/google_web_nl.py +156 -0
- data/core/site-packages/pre_build/VUA_pylib/corpus_reader/google_web_nl.pyc +0 -0
- data/core/site-packages/pre_build/VUA_pylib/io_utils/__init__.py +1 -0
- data/core/site-packages/pre_build/VUA_pylib/io_utils/__init__.pyc +0 -0
- data/core/site-packages/pre_build/VUA_pylib/io_utils/feature_file.py +121 -0
- data/core/site-packages/pre_build/VUA_pylib/io_utils/feature_file.pyc +0 -0
- data/core/site-packages/pre_build/VUA_pylib/lexicon/__init__.py +1 -0
- data/core/site-packages/pre_build/VUA_pylib/lexicon/__init__.pyc +0 -0
- data/core/site-packages/pre_build/VUA_pylib/lexicon/lexicon.py +72 -0
- data/core/site-packages/pre_build/VUA_pylib/lexicon/lexicon.pyc +0 -0
- data/core/site-packages/pre_build/VUKafParserPy-1.0-py2.7.egg-info/PKG-INFO +10 -0
- data/core/site-packages/pre_build/VUKafParserPy-1.0-py2.7.egg-info/SOURCES.txt +7 -0
- data/core/site-packages/pre_build/VUKafParserPy-1.0-py2.7.egg-info/dependency_links.txt +1 -0
- data/core/site-packages/pre_build/VUKafParserPy-1.0-py2.7.egg-info/installed-files.txt +11 -0
- data/core/site-packages/pre_build/VUKafParserPy-1.0-py2.7.egg-info/top_level.txt +1 -0
- data/core/site-packages/pre_build/VUKafParserPy/KafDataObjectsMod.py +165 -0
- data/core/site-packages/pre_build/VUKafParserPy/KafDataObjectsMod.pyc +0 -0
- data/core/site-packages/pre_build/VUKafParserPy/KafParserMod.py +439 -0
- data/core/site-packages/pre_build/VUKafParserPy/KafParserMod.pyc +0 -0
- data/core/site-packages/pre_build/VUKafParserPy/__init__.py +7 -0
- data/core/site-packages/pre_build/VUKafParserPy/__init__.pyc +0 -0
- data/core/vendor/src/crfsuite/AUTHORS +1 -0
- data/core/vendor/src/crfsuite/COPYING +27 -0
- data/core/vendor/src/crfsuite/ChangeLog +103 -0
- data/core/vendor/src/crfsuite/INSTALL +236 -0
- data/core/vendor/src/crfsuite/Makefile.am +19 -0
- data/core/vendor/src/crfsuite/Makefile.in +783 -0
- data/core/vendor/src/crfsuite/README +183 -0
- data/core/vendor/src/crfsuite/aclocal.m4 +9018 -0
- data/core/vendor/src/crfsuite/autogen.sh +38 -0
- data/core/vendor/src/crfsuite/compile +143 -0
- data/core/vendor/src/crfsuite/config.guess +1502 -0
- data/core/vendor/src/crfsuite/config.h.in +198 -0
- data/core/vendor/src/crfsuite/config.sub +1714 -0
- data/core/vendor/src/crfsuite/configure +14273 -0
- data/core/vendor/src/crfsuite/configure.in +149 -0
- data/core/vendor/src/crfsuite/crfsuite.sln +42 -0
- data/core/vendor/src/crfsuite/depcomp +630 -0
- data/core/vendor/src/crfsuite/example/chunking.py +49 -0
- data/core/vendor/src/crfsuite/example/crfutils.py +179 -0
- data/core/vendor/src/crfsuite/example/ner.py +270 -0
- data/core/vendor/src/crfsuite/example/pos.py +78 -0
- data/core/vendor/src/crfsuite/example/template.py +88 -0
- data/core/vendor/src/crfsuite/frontend/Makefile.am +29 -0
- data/core/vendor/src/crfsuite/frontend/Makefile.in +640 -0
- data/core/vendor/src/crfsuite/frontend/dump.c +116 -0
- data/core/vendor/src/crfsuite/frontend/frontend.vcxproj +129 -0
- data/core/vendor/src/crfsuite/frontend/iwa.c +273 -0
- data/core/vendor/src/crfsuite/frontend/iwa.h +65 -0
- data/core/vendor/src/crfsuite/frontend/learn.c +439 -0
- data/core/vendor/src/crfsuite/frontend/main.c +137 -0
- data/core/vendor/src/crfsuite/frontend/option.c +93 -0
- data/core/vendor/src/crfsuite/frontend/option.h +86 -0
- data/core/vendor/src/crfsuite/frontend/readdata.h +38 -0
- data/core/vendor/src/crfsuite/frontend/reader.c +136 -0
- data/core/vendor/src/crfsuite/frontend/tag.c +427 -0
- data/core/vendor/src/crfsuite/genbinary.sh.in +15 -0
- data/core/vendor/src/crfsuite/include/Makefile.am +11 -0
- data/core/vendor/src/crfsuite/include/Makefile.in +461 -0
- data/core/vendor/src/crfsuite/include/crfsuite.h +1063 -0
- data/core/vendor/src/crfsuite/include/crfsuite.hpp +555 -0
- data/core/vendor/src/crfsuite/include/crfsuite_api.hpp +400 -0
- data/core/vendor/src/crfsuite/include/os.h +61 -0
- data/core/vendor/src/crfsuite/install-sh +520 -0
- data/core/vendor/src/crfsuite/lib/cqdb/COPYING +28 -0
- data/core/vendor/src/crfsuite/lib/cqdb/Makefile.am +21 -0
- data/core/vendor/src/crfsuite/lib/cqdb/Makefile.in +549 -0
- data/core/vendor/src/crfsuite/lib/cqdb/cqdb.vcxproj +86 -0
- data/core/vendor/src/crfsuite/lib/cqdb/include/cqdb.h +524 -0
- data/core/vendor/src/crfsuite/lib/cqdb/src/cqdb.c +587 -0
- data/core/vendor/src/crfsuite/lib/cqdb/src/lookup3.c +976 -0
- data/core/vendor/src/crfsuite/lib/crf/Makefile.am +46 -0
- data/core/vendor/src/crfsuite/lib/crf/Makefile.in +721 -0
- data/core/vendor/src/crfsuite/lib/crf/crf.vcxproj +216 -0
- data/core/vendor/src/crfsuite/lib/crf/src/crf1d.h +353 -0
- data/core/vendor/src/crfsuite/lib/crf/src/crf1d_context.c +705 -0
- data/core/vendor/src/crfsuite/lib/crf/src/crf1d_encode.c +943 -0
- data/core/vendor/src/crfsuite/lib/crf/src/crf1d_feature.c +352 -0
- data/core/vendor/src/crfsuite/lib/crf/src/crf1d_model.c +994 -0
- data/core/vendor/src/crfsuite/lib/crf/src/crf1d_tag.c +550 -0
- data/core/vendor/src/crfsuite/lib/crf/src/crfsuite.c +492 -0
- data/core/vendor/src/crfsuite/lib/crf/src/crfsuite_internal.h +236 -0
- data/core/vendor/src/crfsuite/lib/crf/src/crfsuite_train.c +272 -0
- data/core/vendor/src/crfsuite/lib/crf/src/dataset.c +106 -0
- data/core/vendor/src/crfsuite/lib/crf/src/dictionary.c +118 -0
- data/core/vendor/src/crfsuite/lib/crf/src/holdout.c +80 -0
- data/core/vendor/src/crfsuite/lib/crf/src/logging.c +91 -0
- data/core/vendor/src/crfsuite/lib/crf/src/logging.h +48 -0
- data/core/vendor/src/crfsuite/lib/crf/src/params.c +335 -0
- data/core/vendor/src/crfsuite/lib/crf/src/params.h +80 -0
- data/core/vendor/src/crfsuite/lib/crf/src/quark.c +172 -0
- data/core/vendor/src/crfsuite/lib/crf/src/quark.h +46 -0
- data/core/vendor/src/crfsuite/lib/crf/src/rumavl.c +1107 -0
- data/core/vendor/src/crfsuite/lib/crf/src/rumavl.h +160 -0
- data/core/vendor/src/crfsuite/lib/crf/src/train_arow.c +408 -0
- data/core/vendor/src/crfsuite/lib/crf/src/train_averaged_perceptron.c +242 -0
- data/core/vendor/src/crfsuite/lib/crf/src/train_l2sgd.c +507 -0
- data/core/vendor/src/crfsuite/lib/crf/src/train_lbfgs.c +338 -0
- data/core/vendor/src/crfsuite/lib/crf/src/train_passive_aggressive.c +435 -0
- data/core/vendor/src/crfsuite/lib/crf/src/vecmath.h +341 -0
- data/core/vendor/src/crfsuite/ltmain.sh +8413 -0
- data/core/vendor/src/crfsuite/missing +376 -0
- data/core/vendor/src/crfsuite/swig/Makefile.am +13 -0
- data/core/vendor/src/crfsuite/swig/Makefile.in +365 -0
- data/core/vendor/src/crfsuite/swig/crfsuite.cpp +2 -0
- data/core/vendor/src/crfsuite/swig/export.i +32 -0
- data/core/vendor/src/crfsuite/swig/python/README +92 -0
- data/core/vendor/src/crfsuite/swig/python/crfsuite.py +329 -0
- data/core/vendor/src/crfsuite/swig/python/export_wrap.cpp +14355 -0
- data/core/vendor/src/crfsuite/swig/python/export_wrap.h +63 -0
- data/core/vendor/src/crfsuite/swig/python/prepare.sh +9 -0
- data/core/vendor/src/crfsuite/swig/python/sample_tag.py +52 -0
- data/core/vendor/src/crfsuite/swig/python/sample_train.py +68 -0
- data/core/vendor/src/crfsuite/swig/python/setup.py +44 -0
- data/core/vendor/src/crfsuite/win32/stdint.h +679 -0
- data/core/vendor/src/liblbfgs/AUTHORS +1 -0
- data/core/vendor/src/liblbfgs/COPYING +22 -0
- data/core/vendor/src/liblbfgs/ChangeLog +120 -0
- data/core/vendor/src/liblbfgs/INSTALL +231 -0
- data/core/vendor/src/liblbfgs/Makefile.am +10 -0
- data/core/vendor/src/liblbfgs/Makefile.in +638 -0
- data/core/vendor/src/liblbfgs/NEWS +0 -0
- data/core/vendor/src/liblbfgs/README +71 -0
- data/core/vendor/src/liblbfgs/aclocal.m4 +6985 -0
- data/core/vendor/src/liblbfgs/autogen.sh +38 -0
- data/core/vendor/src/liblbfgs/config.guess +1411 -0
- data/core/vendor/src/liblbfgs/config.h.in +64 -0
- data/core/vendor/src/liblbfgs/config.sub +1500 -0
- data/core/vendor/src/liblbfgs/configure +21146 -0
- data/core/vendor/src/liblbfgs/configure.in +107 -0
- data/core/vendor/src/liblbfgs/depcomp +522 -0
- data/core/vendor/src/liblbfgs/include/lbfgs.h +745 -0
- data/core/vendor/src/liblbfgs/install-sh +322 -0
- data/core/vendor/src/liblbfgs/lbfgs.sln +26 -0
- data/core/vendor/src/liblbfgs/lib/Makefile.am +24 -0
- data/core/vendor/src/liblbfgs/lib/Makefile.in +499 -0
- data/core/vendor/src/liblbfgs/lib/arithmetic_ansi.h +133 -0
- data/core/vendor/src/liblbfgs/lib/arithmetic_sse_double.h +294 -0
- data/core/vendor/src/liblbfgs/lib/arithmetic_sse_float.h +298 -0
- data/core/vendor/src/liblbfgs/lib/lbfgs.c +1371 -0
- data/core/vendor/src/liblbfgs/lib/lib.vcxproj +95 -0
- data/core/vendor/src/liblbfgs/ltmain.sh +6426 -0
- data/core/vendor/src/liblbfgs/missing +353 -0
- data/core/vendor/src/liblbfgs/sample/Makefile.am +15 -0
- data/core/vendor/src/liblbfgs/sample/Makefile.in +433 -0
- data/core/vendor/src/liblbfgs/sample/sample.c +81 -0
- data/core/vendor/src/liblbfgs/sample/sample.cpp +126 -0
- data/core/vendor/src/liblbfgs/sample/sample.vcxproj +105 -0
- data/core/vendor/src/svm_light/LICENSE.txt +59 -0
- data/core/vendor/src/svm_light/Makefile +105 -0
- data/core/vendor/src/svm_light/kernel.h +40 -0
- data/core/vendor/src/svm_light/svm_classify.c +197 -0
- data/core/vendor/src/svm_light/svm_common.c +985 -0
- data/core/vendor/src/svm_light/svm_common.h +301 -0
- data/core/vendor/src/svm_light/svm_hideo.c +1062 -0
- data/core/vendor/src/svm_light/svm_learn.c +4147 -0
- data/core/vendor/src/svm_light/svm_learn.h +169 -0
- data/core/vendor/src/svm_light/svm_learn_main.c +397 -0
- data/core/vendor/src/svm_light/svm_loqo.c +211 -0
- data/ext/hack/Rakefile +17 -0
- data/ext/hack/support.rb +88 -0
- data/lib/opener/opinion_detectors/base.rb +112 -0
- data/lib/opener/opinion_detectors/base/version.rb +7 -0
- data/lib/opener/opinion_detectors/configuration_creator.rb +86 -0
- data/lib/opener/opinion_detectors/de.rb +7 -0
- data/lib/opener/opinion_detectors/en.rb +7 -0
- data/lib/opener/opinion_detectors/it.rb +7 -0
- data/lib/opener/opinion_detectors/nl.rb +6 -0
- data/opener-opinion-detector-base.gemspec +35 -0
- data/pre_build_requirements.txt +3 -0
- metadata +374 -0
@@ -0,0 +1,634 @@
|
|
1
|
+
#!/usr/bin/env python
|
2
|
+
|
3
|
+
import argparse
|
4
|
+
import sys
|
5
|
+
import ConfigParser
|
6
|
+
import os
|
7
|
+
import shutil
|
8
|
+
import time
|
9
|
+
|
10
|
+
from subprocess import Popen, PIPE
|
11
|
+
from generate_folds import generate_folds
|
12
|
+
from train import train_all
|
13
|
+
from classify_kaf_naf_file import tag_file_with_opinions
|
14
|
+
from KafNafParserPy import KafNafParser
|
15
|
+
|
16
|
+
|
17
|
+
## CHANGES
|
18
|
+
# 1stApril2014 --> Included evaluation for the opinion miner basic
|
19
|
+
|
20
|
+
default_jar_file='/home/izquierdo/code/triple_evaluation/lib/TripleEvaluation-1.0-jar-with-dependencies.jar'
|
21
|
+
path_to_basic = '/home/izquierdo/opener_repos/opinion-detector-basic/core/opinion_detector_basic_multi.py'
|
22
|
+
|
23
|
+
'''
|
24
|
+
java -Xmx812m -cp $eval_jar_file \
|
25
|
+
vu.tripleevaluation.conversion.ConvertKafToTriples \
|
26
|
+
--kaf-file $1 --opinion
|
27
|
+
'''
|
28
|
+
|
29
|
+
def convert_to_triple(jar_file,input_file):
|
30
|
+
cmd = ['java -Xmx812m']
|
31
|
+
cmd.append('-cp')
|
32
|
+
cmd.append(jar_file)
|
33
|
+
cmd.append('vu.tripleevaluation.conversion.ConvertKafToTriples')
|
34
|
+
cmd.append('--kaf-file')
|
35
|
+
cmd.append(input_file)
|
36
|
+
cmd.append('--opinion')
|
37
|
+
converter = Popen(' '.join(cmd),shell=True, stderr=sys.stderr)
|
38
|
+
converter.wait()
|
39
|
+
|
40
|
+
'''
|
41
|
+
java -Xmx812m -cp $eval_jar_file \
|
42
|
+
vu.tripleevaluation.evaluation.EvaluateTriples \
|
43
|
+
--gold-standard-triples $1 \
|
44
|
+
--system-triples $2 --ignore-relation
|
45
|
+
'''
|
46
|
+
|
47
|
+
## The output is generated in the same folder of the system_triples
|
48
|
+
## with the extension .xls and .log
|
49
|
+
def evaluate_triples(jar_file,gold_triples,system_triples,evaluation_folder):
|
50
|
+
# We need to run it 3 times to
|
51
|
+
# 1) Get the prec/rec for opinion expression
|
52
|
+
# 2) Get the prec/rec for target
|
53
|
+
# 3) Get the prec/red for holder
|
54
|
+
runs = []
|
55
|
+
runs.append(([],'.expression'))
|
56
|
+
runs.append((['--element-second-filter','target'],'.target'))
|
57
|
+
runs.append((['--element-second-filter','holder:'],'.holder')) # The : is not an error!
|
58
|
+
for extra_opts, extension in runs:
|
59
|
+
cmd = ['java -Xmx812m']
|
60
|
+
cmd.append('-cp')
|
61
|
+
cmd.append(jar_file)
|
62
|
+
cmd.append('vu.tripleevaluation.evaluation.EvaluateTriples')
|
63
|
+
cmd.append('--gold-standard-triples')
|
64
|
+
cmd.append(gold_triples)
|
65
|
+
cmd.append('--system-triples')
|
66
|
+
cmd.append(system_triples)
|
67
|
+
cmd.extend(extra_opts)
|
68
|
+
|
69
|
+
|
70
|
+
converter = Popen(' '.join(cmd),shell=True, stderr=PIPE, stdout=PIPE)
|
71
|
+
converter.wait()
|
72
|
+
|
73
|
+
# The output must be on system_triples.trp and .log
|
74
|
+
# Move it to the evaluation folder with the proper extension
|
75
|
+
base_file = os.path.basename(system_triples).replace('.trp','')
|
76
|
+
shutil.move(system_triples+'.xls', evaluation_folder+'/'+base_file+extension)
|
77
|
+
shutil.move(system_triples+'.log', evaluation_folder+'/'+base_file+extension+'.log')
|
78
|
+
|
79
|
+
|
80
|
+
# This function reads all the opinions from the inputfile, maps the opinion labels
|
81
|
+
# with the mapped givenn in config_file:
|
82
|
+
# [valid_opinions]
|
83
|
+
#negative = Negative;StrongNegative
|
84
|
+
#positive = Positive;StrongPositive
|
85
|
+
#
|
86
|
+
# So all Negative,StrongNEgative are mapped to negative
|
87
|
+
# and writes the output to output_file
|
88
|
+
def map_opinion_labels(input_file,output_file,config_file):
|
89
|
+
# Load the mapping from the config_file
|
90
|
+
mapping = {}
|
91
|
+
parser = ConfigParser.ConfigParser()
|
92
|
+
parser.read(config_file)
|
93
|
+
for mapped_opinion, values_in_corpus in parser.items('valid_opinions'):
|
94
|
+
values = [ v for v in values_in_corpus.split(';') if v != '']
|
95
|
+
for v in values:
|
96
|
+
mapping[v] = mapped_opinion
|
97
|
+
del parser
|
98
|
+
##################
|
99
|
+
|
100
|
+
input_kaf = KafNafParser(input_file)
|
101
|
+
remove_these = []
|
102
|
+
for opinion in input_kaf.get_opinions():
|
103
|
+
exp = opinion.get_expression()
|
104
|
+
polarity = exp.get_polarity()
|
105
|
+
if polarity in mapping:
|
106
|
+
mapped_polarity = mapping[polarity]
|
107
|
+
else:
|
108
|
+
opi_id = opinion.get_id()
|
109
|
+
remove_these.append(opi_id)
|
110
|
+
mapped_polarity = polarity
|
111
|
+
|
112
|
+
exp.set_polarity(mapped_polarity)
|
113
|
+
|
114
|
+
for opi_id in remove_these:
|
115
|
+
input_kaf.remove_this_opinion(opi_id)
|
116
|
+
input_kaf.dump(output_file)
|
117
|
+
|
118
|
+
|
119
|
+
def extract_figures(evaluation_file):
|
120
|
+
fic = open(evaluation_file)
|
121
|
+
prec_first = rec_first = prec_second = rec_second = prec_rel = rec_rel = 0
|
122
|
+
in_rel_section = False
|
123
|
+
for line in fic:
|
124
|
+
if line.find('Precision of first elements') != -1:
|
125
|
+
fields = line.strip().split('\t')
|
126
|
+
prec_first = fields[1]
|
127
|
+
if prec_first=='NaN': prec_first='0'
|
128
|
+
elif line.find('Recall of first elements') != -1:
|
129
|
+
fields = line.strip().split('\t')
|
130
|
+
rec_first = fields[1]
|
131
|
+
if rec_first=='NaN': rec_first='0'
|
132
|
+
if line.find('Precision of second elements') != -1:
|
133
|
+
fields = line.strip().split('\t')
|
134
|
+
prec_second = fields[1]
|
135
|
+
if prec_second=='NaN': prec_second='0'
|
136
|
+
elif line.find('Recall of second elements') != -1:
|
137
|
+
fields = line.strip().split('\t')
|
138
|
+
rec_second = fields[1]
|
139
|
+
if rec_second=='NaN': rec_second='0'
|
140
|
+
elif line.find('Results per relation') != -1:
|
141
|
+
in_rel_section = True
|
142
|
+
elif in_rel_section and line.startswith('Total'):
|
143
|
+
fields = line.strip().split('\t')
|
144
|
+
rec_rel = fields[-2]
|
145
|
+
prec_rel = fields[-1]
|
146
|
+
fic.close()
|
147
|
+
return float(prec_first)*100,float(rec_first)*100,float(prec_second)*100,float(rec_second)*100,float(prec_rel),float(rec_rel)
|
148
|
+
|
149
|
+
|
150
|
+
#This functions calls to the
|
151
|
+
def run_basic_version(input_file,output_file):
|
152
|
+
cmd = [path_to_basic]
|
153
|
+
cmd.append('--no-opinion-strength')
|
154
|
+
fin = open(input_file,'r')
|
155
|
+
fout = open(output_file,'w')
|
156
|
+
devnull = open(os.devnull,'wb')
|
157
|
+
basic_opinion_miner = Popen(' '.join(cmd),stdin=fin, stdout=fout,stderr=devnull,shell=True)
|
158
|
+
devnull.close()
|
159
|
+
fin.close()
|
160
|
+
basic_opinion_miner.wait()
|
161
|
+
fout.close()
|
162
|
+
|
163
|
+
|
164
|
+
|
165
|
+
if __name__ == '__main__':
|
166
|
+
#map_opinion_labels('input.kaf','output.kaf','models/hotel_set1_set2_nl_validation/fold_0/config.cfg')
|
167
|
+
#sys.exit(0)
|
168
|
+
|
169
|
+
import logging
|
170
|
+
logging.basicConfig(stream=sys.stderr,format='%(asctime)s - %(levelname)s\n %(message)s', level=logging.DEBUG)
|
171
|
+
|
172
|
+
|
173
|
+
argument_parser = argparse.ArgumentParser(description="Perform fold-cross validation on the opinion mining system")
|
174
|
+
argument_parser.add_argument('-n', type=int, default=10, dest='num_folds',help="Num folds (default 10)")
|
175
|
+
argument_parser.add_argument('--config_file','-f',dest='config_file', required=True, help='Configuration file')
|
176
|
+
argument_parser.add_argument('--eval_jar_file', action='store',dest='eval_jar_file', default=default_jar_file,help='Path to triple evaluation jar file (Default '+default_jar_file+')')
|
177
|
+
argument_parser.add_argument('-id',dest='exp_id',default='unknown',help='Experiment ID')
|
178
|
+
argument_parser.add_argument('-out_folds', dest='out_folds', type=argparse.FileType('wb'),)
|
179
|
+
arguments = argument_parser.parse_args()
|
180
|
+
|
181
|
+
|
182
|
+
##
|
183
|
+
corpus_filename = None
|
184
|
+
output_folder = None
|
185
|
+
##
|
186
|
+
|
187
|
+
## Get the list of KAF files from the config filename
|
188
|
+
my_config = ConfigParser.ConfigParser()
|
189
|
+
my_config.read(arguments.config_file)
|
190
|
+
|
191
|
+
if not my_config.has_option('general', 'filename_training_list'):
|
192
|
+
print>>sys.stderr,'[general] => filename_training_list option missing in ',config_file
|
193
|
+
sys.exit(-1)
|
194
|
+
|
195
|
+
if not my_config.has_option('general','output_folder'):
|
196
|
+
print>>sys.stderr,'[general] => output_folder option missing in ',config_file
|
197
|
+
sys.exit(-1)
|
198
|
+
|
199
|
+
#This is the file with the list of KAF/NAf files for training
|
200
|
+
corpus_filename = my_config.get('general', 'filename_training_list')
|
201
|
+
output_folder = my_config.get('general','output_folder')
|
202
|
+
|
203
|
+
|
204
|
+
# Creating the folds in this way:
|
205
|
+
# output_folder/fold_[0-9]+
|
206
|
+
|
207
|
+
#With this we generate new folds and so on
|
208
|
+
generate_folds(corpus_filename,arguments.num_folds,output_folder)
|
209
|
+
|
210
|
+
|
211
|
+
## With this we copy it from a referencre created split of folds
|
212
|
+
'''
|
213
|
+
reference = '/home/izquierdo/cltl_repos/opinion_miner_deluxe/models/hotel_nl_trainandtest_all'
|
214
|
+
if os.path.exists(output_folder):
|
215
|
+
shutil.rmtree(output_folder)
|
216
|
+
shutil.copytree(reference, output_folder)
|
217
|
+
'''
|
218
|
+
|
219
|
+
|
220
|
+
if arguments.out_folds is not None:
|
221
|
+
arguments.out_folds.write('\\begin{table}\n')
|
222
|
+
arguments.out_folds.write("\\begin{tabular}{c|c|c|c|c|c|c||c|c|c|c}\n")
|
223
|
+
arguments.out_folds.write("\\hline\n")
|
224
|
+
arguments.out_folds.write("Type & \\multicolumn{2}{|c|}{Expression} & \\multicolumn{2}{|c|}{Target} & \\multicolumn{2}{|c||}{Holder} &")
|
225
|
+
arguments.out_folds.write("\\multicolumn{2}{|c|}{Expression-Target} & \\multicolumn{2}{|c|}{Expression-Holder}\n")
|
226
|
+
arguments.out_folds.write("\\hline\n")
|
227
|
+
arguments.out_folds.write("& P & R & P & R & P & R & P & R & P & R \\\\\n")
|
228
|
+
arguments.out_folds.flush()
|
229
|
+
## Process each fold
|
230
|
+
|
231
|
+
#For the deluxe
|
232
|
+
all_files = 0
|
233
|
+
all_p_e = all_r_e = all_p_t = all_r_t = all_p_h = all_r_h = all_p_e_t = all_r_e_t = all_p_e_h = all_r_e_h = 0
|
234
|
+
|
235
|
+
#For the basic
|
236
|
+
all_files_basic = 0
|
237
|
+
all_p_e_basic = all_r_e_basic = all_p_t_basic = all_r_t_basic = all_p_h_basic = all_r_h_basic = 0
|
238
|
+
all_p_e_t_basic = all_r_e_t_basic = all_p_e_h_basic = all_r_e_h_basic = 0
|
239
|
+
for num in range(arguments.num_folds):
|
240
|
+
this_folder = output_folder+'/fold_'+str(num)
|
241
|
+
this_model_folder = this_folder+'/models'
|
242
|
+
this_train_file = this_folder+'/train'
|
243
|
+
this_config = this_folder+'/config.cfg'
|
244
|
+
##this_log = this_folder+'/train.log'
|
245
|
+
|
246
|
+
#Modifying the configuration and saving it to the proper file
|
247
|
+
my_config.set('general','output_folder',this_model_folder)
|
248
|
+
my_config.set('general', 'filename_training_list',this_train_file)
|
249
|
+
fp = open (this_config,'w')
|
250
|
+
my_config.write(fp)
|
251
|
+
fp.close()
|
252
|
+
##################
|
253
|
+
|
254
|
+
# We need to run now the training for this folder
|
255
|
+
#def train_all(file_config):
|
256
|
+
print>>sys.stderr, 'Training'
|
257
|
+
print>>sys.stderr, ' Folder',this_folder,
|
258
|
+
|
259
|
+
this_pid = os.fork()
|
260
|
+
if this_pid == 0:
|
261
|
+
train_all(this_config)
|
262
|
+
sys.exit(0)
|
263
|
+
else:
|
264
|
+
while True:
|
265
|
+
sys.stderr.write('.')
|
266
|
+
sys.stderr.flush()
|
267
|
+
status = os.waitpid(this_pid,os.WNOHANG)
|
268
|
+
if status[0] != 0: break
|
269
|
+
time.sleep(1)
|
270
|
+
|
271
|
+
#####
|
272
|
+
## Do the evaluation
|
273
|
+
####
|
274
|
+
|
275
|
+
|
276
|
+
folder_out_kafs = this_folder+'/output_files'
|
277
|
+
if os.path.exists(folder_out_kafs):
|
278
|
+
shutil.rmtree(folder_out_kafs)
|
279
|
+
os.mkdir(folder_out_kafs)
|
280
|
+
|
281
|
+
folder_out_kafs_basic = this_folder+'/output_files_basic'
|
282
|
+
if os.path.exists(folder_out_kafs_basic):
|
283
|
+
shutil.rmtree(folder_out_kafs_basic)
|
284
|
+
os.mkdir(folder_out_kafs_basic)
|
285
|
+
|
286
|
+
folder_gold_triples = this_folder+'/gold_triples'
|
287
|
+
if os.path.exists(folder_gold_triples): shutil.rmtree(folder_gold_triples)
|
288
|
+
os.mkdir(folder_gold_triples)
|
289
|
+
|
290
|
+
fold_test_corpora = this_folder+'/test'
|
291
|
+
fold_test_corpora_desc = open(fold_test_corpora,'r')
|
292
|
+
|
293
|
+
evaluation_folder = this_folder+'/evaluation'
|
294
|
+
if os.path.exists(evaluation_folder):
|
295
|
+
shutil.rmtree(evaluation_folder)
|
296
|
+
os.mkdir(evaluation_folder)
|
297
|
+
|
298
|
+
evaluation_folder_basic = this_folder+'/evaluation_basic'
|
299
|
+
if os.path.exists(evaluation_folder_basic):
|
300
|
+
shutil.rmtree(evaluation_folder_basic)
|
301
|
+
os.mkdir(evaluation_folder_basic)
|
302
|
+
|
303
|
+
list_triple_files = [] #Pairs (gold,out)
|
304
|
+
list_test_base_files = []
|
305
|
+
for cnt, input_test_file in enumerate(fold_test_corpora_desc):
|
306
|
+
input_test_file = input_test_file.strip()
|
307
|
+
basename_file = os.path.basename(input_test_file)
|
308
|
+
list_test_base_files.append(basename_file)
|
309
|
+
|
310
|
+
# We copy the input KAF file into the gold triple folder doing a mapping
|
311
|
+
# of the opinion labels as the input might contain Positive, Negative, Strong
|
312
|
+
# and the output of the system just positive and neative (internal mapping)
|
313
|
+
|
314
|
+
map_opinion_labels(input_test_file,folder_gold_triples+'/'+basename_file,this_config)
|
315
|
+
print>>sys.stderr, cnt,'| Processing ',basename_file
|
316
|
+
sys.stderr.flush()
|
317
|
+
|
318
|
+
input_test_file = folder_gold_triples+'/'+basename_file
|
319
|
+
gold_triple_file = folder_gold_triples+'/'+basename_file+'.trp'
|
320
|
+
|
321
|
+
|
322
|
+
convert_to_triple(arguments.eval_jar_file, input_test_file)
|
323
|
+
print>>sys.stderr, ' Created triple gold in',os.path.basename(gold_triple_file)
|
324
|
+
sys.stderr.flush()
|
325
|
+
|
326
|
+
##########################
|
327
|
+
#For the deluxe version #
|
328
|
+
##########################
|
329
|
+
output_test_file = folder_out_kafs+'/'+basename_file
|
330
|
+
output_triple_file = folder_out_kafs+'/'+basename_file+'.trp'
|
331
|
+
|
332
|
+
|
333
|
+
tag_file_with_opinions(input_test_file,output_test_file,this_model_folder,remove_existing_opinions=True,include_polarity_strength=False)
|
334
|
+
print>>sys.stderr, ' Classified DELUXE in ',os.path.basename(output_test_file)
|
335
|
+
sys.stderr.flush()
|
336
|
+
|
337
|
+
convert_to_triple(arguments.eval_jar_file,output_test_file)
|
338
|
+
print>>sys.stderr, ' Created triple deluxe system in',os.path.basename(output_triple_file)
|
339
|
+
sys.stderr.flush()
|
340
|
+
|
341
|
+
# Run the evaluation
|
342
|
+
evaluate_triples(arguments.eval_jar_file, gold_triple_file, output_triple_file, evaluation_folder)
|
343
|
+
print>>sys.stderr, ' Evaluated deluxe on',evaluation_folder
|
344
|
+
sys.stderr.flush()
|
345
|
+
#########################
|
346
|
+
|
347
|
+
##########################
|
348
|
+
#For the basic version #
|
349
|
+
##########################
|
350
|
+
output_test_file_basic = folder_out_kafs_basic+'/'+basename_file
|
351
|
+
output_triple_file_basic = folder_out_kafs_basic+'/'+basename_file+'.trp'
|
352
|
+
|
353
|
+
#Call to the opinion miner basic and generate output_test_file_basic
|
354
|
+
print>>sys.stderr,' Start basic version input:',input_test_file,'out',output_test_file_basic
|
355
|
+
run_basic_version(input_test_file,output_test_file_basic)
|
356
|
+
print>>sys.stderr, ' Classified BASIC in ',os.path.basename(output_test_file_basic)
|
357
|
+
sys.stderr.flush()
|
358
|
+
|
359
|
+
convert_to_triple(arguments.eval_jar_file,output_test_file_basic)
|
360
|
+
print>>sys.stderr, ' Created triple basic system in',os.path.basename(output_triple_file_basic)
|
361
|
+
sys.stderr.flush()
|
362
|
+
# Run the evaluation
|
363
|
+
evaluate_triples(arguments.eval_jar_file, gold_triple_file, output_triple_file_basic, evaluation_folder_basic)
|
364
|
+
print>>sys.stderr, ' Evaluated basic on',evaluation_folder_basic
|
365
|
+
sys.stderr.flush()
|
366
|
+
#########################
|
367
|
+
|
368
|
+
|
369
|
+
|
370
|
+
fold_test_corpora_desc.close()
|
371
|
+
|
372
|
+
## Extract the figures from the excel files
|
373
|
+
#DELUXE
|
374
|
+
num_files = 0
|
375
|
+
over_p_e = over_r_e = over_p_t = over_r_t = over_p_h = over_r_h = over_p_e_t = over_r_e_t = over_p_e_h = over_r_e_h = 0
|
376
|
+
for test_file in list_test_base_files:
|
377
|
+
expression_eval_file = evaluation_folder+'/'+test_file+'.expression'
|
378
|
+
target_eval_file = evaluation_folder+'/'+test_file+'.target'
|
379
|
+
holder_eval_file = evaluation_folder+'/'+test_file+'.holder'
|
380
|
+
|
381
|
+
prec_exp, rec_exp, _, _, _, _ = extract_figures(expression_eval_file)
|
382
|
+
_,_, prec_tar,rec_tar, prec_rel_exp_tar, rec_rel_exp_tar = extract_figures(target_eval_file)
|
383
|
+
_,_, prec_hol, rec_hol, prec_rel_exp_hol, rec_rel_exp_hol = extract_figures(holder_eval_file)
|
384
|
+
|
385
|
+
over_p_e += prec_exp
|
386
|
+
over_r_e += rec_exp
|
387
|
+
over_p_t += prec_tar
|
388
|
+
over_r_t += rec_tar
|
389
|
+
over_p_h += prec_hol
|
390
|
+
over_r_h += rec_hol
|
391
|
+
over_p_e_t += prec_rel_exp_tar
|
392
|
+
over_r_e_t += rec_rel_exp_tar
|
393
|
+
over_p_e_h += prec_rel_exp_hol
|
394
|
+
over_r_e_h += rec_rel_exp_hol
|
395
|
+
num_files += 1
|
396
|
+
|
397
|
+
|
398
|
+
#BASIC
|
399
|
+
num_files_basic = 0
|
400
|
+
over_p_e_basic = over_r_e_basic = over_p_t_basic = over_r_t_basic = over_p_h_basic = 0
|
401
|
+
over_r_h_basic = over_p_e_t_basic = over_r_e_t_basic = over_p_e_h_basic = over_r_e_h_basic = 0
|
402
|
+
for test_file in list_test_base_files:
|
403
|
+
expression_eval_file = evaluation_folder_basic+'/'+test_file+'.expression'
|
404
|
+
target_eval_file = evaluation_folder_basic+'/'+test_file+'.target'
|
405
|
+
holder_eval_file = evaluation_folder_basic+'/'+test_file+'.holder'
|
406
|
+
|
407
|
+
prec_exp, rec_exp, _, _, _, _ = extract_figures(expression_eval_file)
|
408
|
+
_,_, prec_tar,rec_tar, prec_rel_exp_tar, rec_rel_exp_tar = extract_figures(target_eval_file)
|
409
|
+
_,_, prec_hol, rec_hol, prec_rel_exp_hol, rec_rel_exp_hol = extract_figures(holder_eval_file)
|
410
|
+
|
411
|
+
over_p_e_basic += prec_exp
|
412
|
+
over_r_e_basic += rec_exp
|
413
|
+
over_p_t_basic += prec_tar
|
414
|
+
over_r_t_basic += rec_tar
|
415
|
+
over_p_h_basic += prec_hol
|
416
|
+
over_r_h_basic += rec_hol
|
417
|
+
over_p_e_t_basic += prec_rel_exp_tar
|
418
|
+
over_r_e_t_basic += rec_rel_exp_tar
|
419
|
+
over_p_e_h_basic += prec_rel_exp_hol
|
420
|
+
over_r_e_h_basic += rec_rel_exp_hol
|
421
|
+
num_files_basic += 1
|
422
|
+
|
423
|
+
if arguments.out_folds is not None:
|
424
|
+
arguments.out_folds.write("%s & %.2f & %.2f & %.2f & %.2f & %.2f & %.2f & %.2f & %.2f & %.2f & %.2f\\\\\n" % ("Deluxe Fold "+str(num),
|
425
|
+
over_p_e*1.0/num_files,
|
426
|
+
over_r_e*1.0/num_files,
|
427
|
+
over_p_t*1.0/num_files,
|
428
|
+
over_r_t*1.0/num_files,
|
429
|
+
over_p_h*1.0/num_files,
|
430
|
+
over_r_h*1.0/num_files,
|
431
|
+
over_p_e_t*1.0/num_files,
|
432
|
+
over_r_e_t*1.0/num_files,
|
433
|
+
over_p_e_h*1.0/num_files,
|
434
|
+
over_r_e_h*1.0/num_files))
|
435
|
+
arguments.out_folds.write("%s & %.2f & %.2f & %.2f & %.2f & %.2f & %.2f & %.2f & %.2f & %.2f & %.2f\\\\\n" % ("Basic Fold "+str(num),
|
436
|
+
over_p_e_basic*1.0/num_files_basic,
|
437
|
+
over_r_e_basic*1.0/num_files_basic,
|
438
|
+
over_p_t_basic*1.0/num_files_basic,
|
439
|
+
over_r_t_basic*1.0/num_files_basic,
|
440
|
+
over_p_h_basic*1.0/num_files_basic,
|
441
|
+
over_r_h_basic*1.0/num_files_basic,
|
442
|
+
over_p_e_t_basic*1.0/num_files_basic,
|
443
|
+
over_r_e_t_basic*1.0/num_files_basic,
|
444
|
+
over_p_e_h_basic*1.0/num_files_basic,
|
445
|
+
over_r_e_h_basic*1.0/num_files_basic
|
446
|
+
))
|
447
|
+
arguments.out_folds.flush()
|
448
|
+
|
449
|
+
|
450
|
+
|
451
|
+
print>>sys.stderr
|
452
|
+
print>>sys.stderr, '#'*30
|
453
|
+
print>>sys.stderr, 'OVERALL DELUXE RESULTS FOR FOLD NUM',num
|
454
|
+
print>>sys.stderr, 'Num files:',num_files
|
455
|
+
print>>sys.stderr, ' Expression:'
|
456
|
+
print>>sys.stderr, ' Precision:',over_p_e*1.0/num_files
|
457
|
+
print>>sys.stderr, ' Recall: ',over_r_e*1.0/num_files
|
458
|
+
print>>sys.stderr, ' Target:'
|
459
|
+
print>>sys.stderr, ' Precision:',over_p_t*1.0/num_files
|
460
|
+
print>>sys.stderr, ' Recall: ',over_r_t*1.0/num_files
|
461
|
+
print>>sys.stderr, ' Holder:'
|
462
|
+
print>>sys.stderr, ' Precision:',over_p_h*1.0/num_files
|
463
|
+
print>>sys.stderr, ' Recall: ',over_r_h*1.0/num_files
|
464
|
+
print>>sys.stderr, ' Relation:'
|
465
|
+
print>>sys.stderr, ' Exp-Tar'
|
466
|
+
print>>sys.stderr, ' Prec:',over_p_e_t*1.0/num_files
|
467
|
+
print>>sys.stderr, ' Rec: ',over_r_e_t*1.0/num_files
|
468
|
+
print>>sys.stderr, ' Exp-Hol'
|
469
|
+
print>>sys.stderr, ' Prec:',over_p_e_h*1.0/num_files
|
470
|
+
print>>sys.stderr, ' Rec: ',over_r_e_h*1.0/num_files
|
471
|
+
print>>sys.stderr, '#'*30
|
472
|
+
all_files += num_files
|
473
|
+
all_p_e +=over_p_e
|
474
|
+
all_r_e += over_r_e
|
475
|
+
all_p_t += over_p_t
|
476
|
+
all_r_t += over_r_t
|
477
|
+
all_p_h += over_p_h
|
478
|
+
all_r_h += over_r_h
|
479
|
+
all_p_e_t += over_p_e_t
|
480
|
+
all_r_e_t += over_r_e_t
|
481
|
+
all_p_e_h += over_p_e_h
|
482
|
+
all_r_e_h += over_r_e_h
|
483
|
+
|
484
|
+
print>>sys.stderr
|
485
|
+
print>>sys.stderr, '#'*30
|
486
|
+
print>>sys.stderr, 'OVERALL BASIC RESULTS FOR FOLD NUM',num
|
487
|
+
print>>sys.stderr, 'Num files:',num_files_basic
|
488
|
+
print>>sys.stderr, ' Expression:'
|
489
|
+
print>>sys.stderr, ' Precision:',over_p_e_basic*1.0/num_files_basic
|
490
|
+
print>>sys.stderr, ' Recall: ',over_r_e_basic*1.0/num_files_basic
|
491
|
+
print>>sys.stderr, ' Target:'
|
492
|
+
print>>sys.stderr, ' Precision:',over_p_t_basic*1.0/num_files_basic
|
493
|
+
print>>sys.stderr, ' Recall: ',over_r_t_basic*1.0/num_files_basic
|
494
|
+
print>>sys.stderr, ' Holder:'
|
495
|
+
print>>sys.stderr, ' Precision:',over_p_h_basic*1.0/num_files_basic
|
496
|
+
print>>sys.stderr, ' Recall: ',over_r_h_basic*1.0/num_files_basic
|
497
|
+
print>>sys.stderr, ' Relation:'
|
498
|
+
print>>sys.stderr, ' Exp-Tar'
|
499
|
+
print>>sys.stderr, ' Prec:',over_p_e_t_basic*1.0/num_files_basic
|
500
|
+
print>>sys.stderr, ' Rec: ',over_r_e_t_basic*1.0/num_files_basic
|
501
|
+
print>>sys.stderr, ' Exp-Hol'
|
502
|
+
print>>sys.stderr, ' Prec:',over_p_e_h_basic*1.0/num_files_basic
|
503
|
+
print>>sys.stderr, ' Rec: ',over_r_e_h_basic*1.0/num_files_basic
|
504
|
+
print>>sys.stderr, '#'*30
|
505
|
+
print>>sys.stderr
|
506
|
+
all_files_basic += num_files_basic
|
507
|
+
all_p_e_basic +=over_p_e_basic
|
508
|
+
all_r_e_basic += over_r_e_basic
|
509
|
+
all_p_t_basic += over_p_t_basic
|
510
|
+
all_r_t_basic += over_r_t_basic
|
511
|
+
all_p_h_basic += over_p_h_basic
|
512
|
+
all_r_h_basic += over_r_h_basic
|
513
|
+
all_p_e_t_basic += over_p_e_t_basic
|
514
|
+
all_r_e_t_basic += over_r_e_t_basic
|
515
|
+
all_p_e_h_basic += over_p_e_h_basic
|
516
|
+
all_r_e_h_basic += over_r_e_h_basic
|
517
|
+
##break ## JUST FOR ONE
|
518
|
+
|
519
|
+
|
520
|
+
|
521
|
+
## NEXT FOLD!!!
|
522
|
+
print>>sys.stderr
|
523
|
+
print>>sys.stderr, '#'*30
|
524
|
+
print>>sys.stderr, 'FINAL DELUXE OVERALL RESULTS FOR ALL FOLDS (microaverage)'
|
525
|
+
print>>sys.stderr, 'Total Num files:',all_files
|
526
|
+
print>>sys.stderr, 'Total num folds:',num
|
527
|
+
print>>sys.stderr, ' Expression:'
|
528
|
+
print>>sys.stderr, ' Precision:',all_p_e*1.0/all_files
|
529
|
+
print>>sys.stderr, ' Recall: ',all_r_e*1.0/all_files
|
530
|
+
print>>sys.stderr, ' Target:'
|
531
|
+
print>>sys.stderr, ' Precision:',all_p_t*1.0/all_files
|
532
|
+
print>>sys.stderr, ' Recall: ',all_r_t*1.0/all_files
|
533
|
+
print>>sys.stderr, ' Holder:'
|
534
|
+
print>>sys.stderr, ' Precision:',all_p_h*1.0/all_files
|
535
|
+
print>>sys.stderr, ' Recall: ',all_r_h*1.0/all_files
|
536
|
+
print>>sys.stderr, ' Relation:'
|
537
|
+
print>>sys.stderr, ' Exp-Tar'
|
538
|
+
print>>sys.stderr, ' Prec:',all_p_e_t*1.0/all_files
|
539
|
+
print>>sys.stderr, ' Rec: ',all_r_e_t*1.0/all_files
|
540
|
+
print>>sys.stderr, ' Exp-Hol'
|
541
|
+
print>>sys.stderr, ' Prec:',all_p_e_h*1.0/all_files
|
542
|
+
print>>sys.stderr, ' Rec: ',all_r_e_h*1.0/all_files
|
543
|
+
print>>sys.stderr, '#'*30
|
544
|
+
print>>sys.stderr
|
545
|
+
|
546
|
+
print>>sys.stderr, 'LATEX DELUXE'
|
547
|
+
print>>sys.stderr, '\\begin{table}'
|
548
|
+
print>>sys.stderr, '\\begin{tabular}{c|c|c|}'
|
549
|
+
print>>sys.stderr, '\\hline'
|
550
|
+
print>>sys.stderr, 'Type & Precision & Recall\\\\'
|
551
|
+
print>>sys.stderr, '\\hline'
|
552
|
+
print>>sys.stderr, 'Expression & %.2f & %.2f\\\\' % (all_p_e*1.0/all_files,all_r_e*1.0/all_files)
|
553
|
+
print>>sys.stderr, 'Target & %.2f & %.2f \\\\' % (all_p_t*1.0/all_files,all_r_t*1.0/all_files)
|
554
|
+
print>>sys.stderr, 'Holder & %.2f & %.2f \\\\' % (all_p_h*1.0/all_files,all_r_h*1.0/all_files)
|
555
|
+
print>>sys.stderr, '\\hline'
|
556
|
+
print>>sys.stderr, 'Exp-Tar & %.2f & %.2f\\\\' % (all_p_e_t*1.0/all_files,all_r_e_t*1.0/all_files)
|
557
|
+
print>>sys.stderr, 'Exp-Hol & %.2f & %.2f\\\\' % (all_p_e_h*1.0/all_files,all_r_e_h*1.0/all_files)
|
558
|
+
print>>sys.stderr, '\\hline'
|
559
|
+
print>>sys.stderr, '\\end{tabular}'
|
560
|
+
print>>sys.stderr, '\\caption{Caption here}'
|
561
|
+
print>>sys.stderr, '\\end{table}'
|
562
|
+
|
563
|
+
|
564
|
+
print>>sys.stderr
|
565
|
+
print>>sys.stderr, '#'*30
|
566
|
+
print>>sys.stderr, 'FINAL OVERALL BASIC RESULTS FOR ALL FOLDS (microaverage)'
|
567
|
+
print>>sys.stderr, 'Total Num files:',all_files_basic
|
568
|
+
print>>sys.stderr, 'Total num folds:',num
|
569
|
+
print>>sys.stderr, ' Expression:'
|
570
|
+
print>>sys.stderr, ' Precision:',all_p_e_basic*1.0/all_files_basic
|
571
|
+
print>>sys.stderr, ' Recall: ',all_r_e_basic*1.0/all_files_basic
|
572
|
+
print>>sys.stderr, ' Target:'
|
573
|
+
print>>sys.stderr, ' Precision:',all_p_t_basic*1.0/all_files_basic
|
574
|
+
print>>sys.stderr, ' Recall: ',all_r_t_basic*1.0/all_files_basic
|
575
|
+
print>>sys.stderr, ' Holder:'
|
576
|
+
print>>sys.stderr, ' Precision:',all_p_h_basic*1.0/all_files_basic
|
577
|
+
print>>sys.stderr, ' Recall: ',all_r_h_basic*1.0/all_files_basic
|
578
|
+
print>>sys.stderr, ' Relation:'
|
579
|
+
print>>sys.stderr, ' Exp-Tar'
|
580
|
+
print>>sys.stderr, ' Prec:',all_p_e_t_basic*1.0/all_files_basic
|
581
|
+
print>>sys.stderr, ' Rec: ',all_r_e_t_basic*1.0/all_files_basic
|
582
|
+
print>>sys.stderr, ' Exp-Hol'
|
583
|
+
print>>sys.stderr, ' Prec:',all_p_e_h_basic*1.0/all_files_basic
|
584
|
+
print>>sys.stderr, ' Rec: ',all_r_e_h_basic*1.0/all_files_basic
|
585
|
+
print>>sys.stderr, '#'*30
|
586
|
+
print>>sys.stderr
|
587
|
+
|
588
|
+
print>>sys.stderr, 'LATEX BASIC'
|
589
|
+
print>>sys.stderr, '\\begin{table}'
|
590
|
+
print>>sys.stderr, '\\begin{tabular}{c|c|c|}'
|
591
|
+
print>>sys.stderr, '\\hline'
|
592
|
+
print>>sys.stderr, 'Type & Precision & Recall\\\\'
|
593
|
+
print>>sys.stderr, '\\hline'
|
594
|
+
print>>sys.stderr, 'Expression & %.2f & %.2f\\\\' % (all_p_e_basic*1.0/all_files_basic,all_r_e_basic*1.0/all_files_basic)
|
595
|
+
print>>sys.stderr, 'Target & %.2f & %.2f \\\\' % (all_p_t_basic*1.0/all_files_basic,all_r_t_basic*1.0/all_files_basic)
|
596
|
+
print>>sys.stderr, 'Holder & %.2f & %.2f \\\\' % (all_p_h_basic*1.0/all_files_basic,all_r_h_basic*1.0/all_files_basic)
|
597
|
+
print>>sys.stderr, '\\hline'
|
598
|
+
print>>sys.stderr, 'Exp-Tar & %.2f & %.2f\\\\' % (all_p_e_t_basic*1.0/all_files_basic,all_r_e_t_basic*1.0/all_files_basic)
|
599
|
+
print>>sys.stderr, 'Exp-Hol & %.2f & %.2f\\\\' % (all_p_e_h_basic*1.0/all_files_basic,all_r_e_h_basic*1.0/all_files_basic)
|
600
|
+
print>>sys.stderr, '\\hline'
|
601
|
+
print>>sys.stderr, '\\end{tabular}'
|
602
|
+
print>>sys.stderr, '\\caption{Caption here}'
|
603
|
+
print>>sys.stderr, '\\end{table}'
|
604
|
+
|
605
|
+
|
606
|
+
if arguments.out_folds is not None:
|
607
|
+
arguments.out_folds.write("\\end{tabular}\n")
|
608
|
+
arguments.out_folds.write("\\end{table}\n")
|
609
|
+
arguments.out_folds.close()
|
610
|
+
|
611
|
+
##Output for Latex
|
612
|
+
print "%s Deluxe & %.2f & %.2f & %.2f & %.2f & %.2f & %.2f & %.2f & %.2f & %.2f & %.2f\\\\" % (arguments.exp_id,
|
613
|
+
all_p_e * 1.0 /all_files,all_r_e *1.0/all_files,
|
614
|
+
all_p_t * 1.0 /all_files,all_r_t *1.0/all_files,
|
615
|
+
all_p_h * 1.0 /all_files,all_r_h *1.0/all_files,
|
616
|
+
all_p_e_t * 1.0/all_files,all_r_e_t*1.0/all_files,
|
617
|
+
all_p_e_h * 1.0/all_files,all_r_e_h*1.0/all_files
|
618
|
+
)
|
619
|
+
print "%s Basic & %.2f & %.2f & %.2f & %.2f & %.2f & %.2f & %.2f & %.2f & %.2f & %.2f\\\\" % (arguments.exp_id,
|
620
|
+
all_p_e_basic *1.0/all_files_basic,all_r_e_basic *1.0/all_files_basic,
|
621
|
+
all_p_t_basic *1.0/all_files_basic,all_r_t_basic *1.0/all_files_basic,
|
622
|
+
all_p_h_basic *1.0/all_files_basic,all_r_h_basic *1.0/all_files_basic,
|
623
|
+
all_p_e_t_basic*1.0/all_files_basic,all_r_e_t_basic*1.0/all_files_basic,
|
624
|
+
all_p_e_h_basic*1.0/all_files_basic,all_r_e_h_basic*1.0/all_files_basic
|
625
|
+
)
|
626
|
+
|
627
|
+
|
628
|
+
|
629
|
+
|
630
|
+
|
631
|
+
|
632
|
+
|
633
|
+
sys.exit(0)
|
634
|
+
|