evaluation.py
11.5 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
#-*- coding:utf-8 -*-
import codecs
import datetime
import itertools
import os
import random
import time
from django.core.management.base import BaseCommand
from normalization import normalize
from verification.duckduckgo import DuckDuckGo
from webapp.models import Expression, Meaning, Source
from settings import PROJECT_PATH
SOURCE = '1M'
MEANINGS_PATH = os.path.join(PROJECT_PATH, 'data', 'meanings_eval-%s-%s.csv' %
(SOURCE, datetime.datetime.now().strftime('%Y%m%d')))
EXPRESSIONS_PATH = os.path.join(PROJECT_PATH, 'data', 'expressions_eval-%s-%s.csv' %
(SOURCE, datetime.datetime.now().strftime('%Y%m%d')))
NORMALIZATION_PATH = os.path.join(PROJECT_PATH, 'data', 'normalization_eval-%s-%s.csv' %
(SOURCE, datetime.datetime.now().strftime('%Y%m%d')))
AMBIG_NORMALIZATION_PATH = os.path.join(PROJECT_PATH, 'data', 'ambig_normalization_eval-%s-%s.csv' %
(SOURCE, datetime.datetime.now().strftime('%Y%m%d')))
MIN_HITS = [3, 5, 10, 15, 20]
SAMPLE = 200
class Command(BaseCommand):
help = 'Evaluate normalization and extraction..'
def handle(self, *args, **options):
evaluate_meanings(MEANINGS_PATH)
evaluate_expressions(EXPRESSIONS_PATH)
evaluate_normalization(NORMALIZATION_PATH, False)
evaluate_normalization(AMBIG_NORMALIZATION_PATH, True)
def evaluate_meanings(eval_path):
csv_file = codecs.open(eval_path, 'wt', 'utf-8')
for source in Source.objects.filter(key=SOURCE):
print (source)
csv_file.write(u'\n%s\n' % source.name)
csv_file.write(u'ID\tHasło\tDefinicja')
for min_hits in MIN_HITS:
csv_file.write('\t%d' % min_hits)
csv_file.write(u'\tCzłowiek\n')
meanings = Meaning.objects.filter(expressions__link__source=source).all()
random_pks_list = []
evaluation = {}
while len(random_pks_list) < SAMPLE and meanings.count() > 0:
random_meaning = random.choice(meanings)
if random_meaning.expressions.filter(is_catchword=True).exists() and random_meaning.expressions > 1:
meanings = meanings.exclude(pk=random_meaning.pk)
random_pks_list.append(random_meaning.pk)
evaluation[random_meaning.pk] = {'meaning': random_meaning}
for min_hits in MIN_HITS:
evaluation[random_meaning.pk][min_hits] = -1
for min_hits in MIN_HITS:
random_meanings = Meaning.objects.filter(pk__in=random_pks_list)
while random_meanings:
random_meanings = check_meanings(evaluation, min_hits, random_meanings)
write_meanings_evaluation(evaluation, csv_file)
csv_file.close()
def check_meanings(evaluation, min_hits, meanings):
check_again_meanings = []
duckduckgo = DuckDuckGo(min_hits)
for meaning in meanings:
for catchword in meaning.expressions.filter(is_catchword=True):
for expression in meaning.expressions.exclude(is_catchword=True):
time.sleep(random.uniform(0.5, 3.0))
try:
if not duckduckgo.check_entry(catchword, expression):
print ('Erase:\t', catchword.orth_text, u'\t-->\t', expression.orth_text)
evaluation[meaning.pk]['definition'] = u'%s\t%s' % (catchword.orth_text,
expression.orth_text)
evaluation[meaning.pk][min_hits] = 0
else:
print ('OK:\t', catchword.orth_text, u'\t-->\t', expression.orth_text)
evaluation[meaning.pk]['definition'] = u'%s\t%s' % (catchword.orth_text,
expression.orth_text)
evaluation[meaning.pk][min_hits] = 1
except RuntimeError:
print ('Try again:\t', catchword.orth_text, u'\t-->\t', expression.orth_text)
check_again_meanings.append(meaning)
break
return check_again_meanings
def write_meanings_evaluation(evaluation, csv_file):
for mng_pk in evaluation:
mng_eval = evaluation[mng_pk]
csv_file.write(u'%d\t%s' % (mng_pk, mng_eval['definition']))
for min_hits in MIN_HITS:
csv_file.write('\t%d' % mng_eval[min_hits])
csv_file.write(u'\n')
def evaluate_expressions(eval_path):
csv_file = codecs.open(eval_path, 'wt', 'utf-8')
for source in Source.objects.filter(key=SOURCE):
print (source)
csv_file.write(u'\n%s\n' % source.name)
csv_file.write(u'ID\tWyrażenie')
for min_hits in MIN_HITS:
csv_file.write('\t%d' % min_hits)
csv_file.write(u'\tCzłowiek\n')
expressions = Expression.objects.filter(link__source=source).all()
random_pks_list = []
evaluation = {}
while len(random_pks_list) < SAMPLE and expressions.count() > 0:
random_expression = random.choice(expressions)
if random_expression.segments.count() > 1:
expressions = expressions.exclude(pk=random_expression.pk)
random_pks_list.append(random_expression.pk)
evaluation[random_expression.pk] = {'expression': random_expression}
for min_hits in MIN_HITS:
evaluation[random_expression.pk][min_hits] = -1
for min_hits in MIN_HITS:
random_expressions = Expression.objects.filter(pk__in=random_pks_list)
while random_expressions:
random_expressions = check_expressions(evaluation, min_hits, random_expressions)
write_expressions_evaluation(evaluation, csv_file)
csv_file.close()
def check_expressions(evaluation, min_hits, random_expressions):
check_again_expressions = []
duckduckgo = DuckDuckGo(min_hits)
for expression in random_expressions:
time.sleep(random.uniform(0.5, 3.0))
try:
if not duckduckgo.check_expression(expression):
print ('Erase:\t', expression.orth_text)
evaluation[expression.pk]['expression'] = u'%s' % expression.orth_text
evaluation[expression.pk][min_hits] = 0
else:
print ('OK:\t', expression.orth_text)
evaluation[expression.pk]['expression'] = u'%s' % expression.orth_text
evaluation[expression.pk][min_hits] = 1
except RuntimeError:
print ('Try again:\t', expression.orth_text)
check_again_expressions.append(expression)
break
return check_again_expressions
def write_expressions_evaluation(evaluation, csv_file):
for expr_pk in evaluation:
expr_eval = evaluation[expr_pk]
csv_file.write(u'%d\t%s' % (expr_pk, expr_eval['expression']))
for min_hits in MIN_HITS:
csv_file.write('\t%d' % expr_eval[min_hits])
csv_file.write(u'\n')
def evaluate_normalization(eval_path, ambiguity):
csv_file = codecs.open(eval_path, 'wt', 'utf-8')
for source in Source.objects.filter(key=SOURCE):
print (source)
csv_file.write(u'\n%s\n' % source.name)
csv_file.write(u'ID\tWyrażenie bazowe\tNormalizacja')
for min_hits in MIN_HITS:
csv_file.write('\t%d' % min_hits)
csv_file.write(u'\tCzłowiek\n')
expressions = Expression.objects.filter(link__source=source).all()
random_pks_list = []
evaluation = {}
while len(random_pks_list) < SAMPLE and expressions.count() > 0:
random_expression = random.choice(expressions)
if (random_expression.segments.count() > 1 and
len(get_normalized_expression(random_expression, ambiguity)) > 0):
expressions = expressions.exclude(pk=random_expression.pk)
random_pks_list.append(random_expression.pk)
evaluation[random_expression.pk] = {'expression': random_expression}
for min_hits in MIN_HITS:
evaluation[random_expression.pk][min_hits] = -1
for min_hits in MIN_HITS:
random_expressions = Expression.objects.filter(pk__in=random_pks_list)
while random_expressions:
random_expressions = check_normalizations(evaluation, min_hits, random_expressions, ambiguity)
write_normalizations_evaluation(evaluation, csv_file)
csv_file.close()
def check_normalizations(evaluation, min_hits, random_expressions, ambiguity):
check_again_expressions = []
duckduckgo = DuckDuckGo(min_hits)
for expression in random_expressions:
for normalized_expr in get_normalized_expression(expression, ambiguity):
time.sleep(random.uniform(0.5, 3.0))
try:
if not duckduckgo.check_expression(expression, normalized_expr):
print ('Erase:\t', expression.orth_text, '!!', normalized_expr)
evaluation[expression.pk]['expression'] = u'%s\t%s' % (expression.orth_text, normalized_expr)
evaluation[expression.pk][min_hits] = 0
else:
print ('OK:\t', expression.orth_text, '!!', normalized_expr)
evaluation[expression.pk]['expression'] = u'%s\t%s' % (expression.orth_text, normalized_expr)
evaluation[expression.pk][min_hits] = 1
except RuntimeError:
print ('Try again:\t', expression.orth_text, '!!', normalized_expr)
check_again_expressions.append(expression)
break
return check_again_expressions
def get_normalized_expression(expression, ambiguity):
normalized_expressions = []
no_spaced_orig_expr = expression.orth_text.replace(' ', '')
if expression.segments.count() > 1:
possible_forms = generate_forms(expression, ambiguity)
if len(possible_forms) > 1:
for form in possible_forms:
normalized_expression = normalize.get_normalized_expr_text(expression, form)
no_spaced_normalized_expr = normalized_expression.replace(' ', '')
if no_spaced_normalized_expr == no_spaced_orig_expr:
continue
normalized_expressions.append(form)
return set(normalized_expressions)
def generate_forms(expression, ambiguity):
segments = []
expr_segments = expression.segments.order_by('position_in_expr')
for seg in expr_segments:
seg_lus = normalize.get_lus(seg.base)
if seg.is_head and (seg_lus.count() == 1 or ambiguity):
head_equivalents = normalize.get_head_equivalents(seg)
segments.append(head_equivalents)
elif seg_lus.count() == 1 or ambiguity:
equivalents = normalize.get_synonymic_equivalents(seg)
segments.append(equivalents)
else:
segments.append([seg.orth])
generated_expressions = itertools.product(*segments)
generated_expressions = list(generated_expressions)
pariciple_expressions = normalize.create_pariciple_expressions(generated_expressions, expression)
generated_expressions.extend(pariciple_expressions)
return generated_expressions
def write_normalizations_evaluation(evaluation, csv_file):
for expr_pk in evaluation:
expr_eval = evaluation[expr_pk]
csv_file.write(u'%d\t%s' % (expr_pk, expr_eval['expression']))
for min_hits in MIN_HITS:
csv_file.write('\t%d' % expr_eval[min_hits])
csv_file.write(u'\n')