normalize.py
6.82 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
import itertools
import settings
import nkjp
from webapp.models import Segment
from wordnet.models import LexicalUnit, Relation
# dodac inwersje, np. intrygant kulkowy >> kulkowy intrygant !!!
def expressions(expressions):
orth_expressions = []
nkjp_connection = nkjp.connect('nkjp1800x')
nkjp_connection = None
for expr in expressions.order_by('text'):
orth_expressions.extend(expression(nkjp_connection, expr))
if expr.orth_text not in orth_expressions:
orth_expressions.append(expr.orth_text)
nkjp_connection.close()
return orth_expressions
def expression(nkjp_connection, expression):
normalized_expressions = []
if expression.segments.count() > 1 and expression.segments.count() < 5:
possible_forms = generate_forms(expression)
for form in possible_forms:
normalized_expression = get_normalized_expr_text(expression, form)
normalized_expression_query = get_normalized_expr_query(expression, form)
if nkjp.contains(nkjp_connection, normalized_expression_query):
normalized_expressions.append(normalized_expression)
# tutaj sie bedzie jeszcze bruzdzic --> stawic sie np.
#normalize_verbs(expression)
# normalized_expressions.extend(normalize_head(nkjp_connection, expression))
return set(normalized_expressions)
def generate_forms(expression):
segments = []
for seg in expression.segments.order_by('position_in_expr'):
if seg.is_verb() and seg.orth != u'jest':
verb_equivalents = get_verb_equivalents(seg)
segments.append(verb_equivalents)
elif seg.is_head:# and get_lus(seg.base).count() == 1:
head_equivalents = get_head_equivalents(seg)
segments.append(head_equivalents)
elif seg.is_noun() and get_lus(seg.base).count() == 1:
verb_equivalents = get_verb_equivalents(seg)
segments.append(verb_equivalents)
else:
segments.append([seg.orth])
return list(itertools.product(*segments))
def get_verb_equivalents(seg):
all_equivalents = []
all_equivalents.extend(get_synonyms(seg.base))
inflected_equivalents = [seg.orth]
for equivalent in all_equivalents:
forms = segment(seg, equivalent)
inflected_equivalents.extend(forms)
return list(set(inflected_equivalents))
def get_head_equivalents(seg):
all_equivalents = []
all_equivalents.extend(get_synonyms(seg.base))
all_equivalents.extend(get_by_relation(seg.base, 1,
Relation.objects.get(parent=None, name='hiperonimia')))
inflected_equivalents = [seg.orth]
for equivalent in all_equivalents:
forms = segment(seg, equivalent)
inflected_equivalents.extend(forms)
return list(set(inflected_equivalents))
# def normalize_head(nkjp_connection, expression):
# normalized_expressions = []
# try:
# head = expression.segments.get(is_head=True)
# all_equivalents = []
# head_lus = get_lus(head.base)
# all_equivalents.extend(get_synonyms(head.base))
# all_equivalents.extend(get_by_relation(head.base, 1,
# Relation.objects.get(parent=None, name='hiperonimia')))
# for synonym in all_equivalents:
# synonym_forms = segment(head, synonym)
# for form in synonym_forms:
# normalized_expression = get_normalized_expr_text(expression, form)
# normalized_expression_query = get_normalized_expr_query(expression, form)
# if head_lus.count() == 1 or nkjp.contains(nkjp_connection, normalized_expression_query):
# normalized_expressions.append(normalized_expression)
# except Segment.DoesNotExist:
# pass
#
# return set(normalized_expressions)
def get_synonyms(base):
synonyms = []
synsets = get_synsets(base)
for synset in synsets:
for lu in synset.lus.all():
if (base != lu.base and len(lu.base.split()) == 1
and lu.base not in synonyms):
synonyms.append(lu.base)
return synonyms
def get_synsets(base):
synsets = []
lus = get_lus(base)
for lu in lus:
if lu.synset not in synsets:
synsets.append(lu.synset)
return synsets
def get_lus(base):
return LexicalUnit.objects.filter(base=base)
def get_by_relation(base, max_depth, relation):
related_bases = []
source_synsets = get_synsets(base)
related_synsets = get_related_synsets(source_synsets, relation, max_depth)
for synset in related_synsets:
if synset not in source_synsets:
for lu in synset.lus.all():
if (base != lu.base and len(lu.base.split()) == 1
and lu.base not in related_bases):
related_bases.append(lu.base)
return related_bases
def get_related_synsets(sources, relation, max_depth):
related_synsets = []
for source in sources:
related_synsets.extend(get_related(source, relation, max_depth, 0))
return related_synsets
def get_related(source, relation, max_depth, depth):
depth += 1
visited = [source]
if depth > max_depth:
return visited
links = source.targets.filter(relation=relation)
for lid, link in enumerate(links):
visited.extend(get_related(link.parent, relation, max_depth, depth))
return visited
def get_normalized_expr_text(expression, form_segments):
expr = ''
expr_segments = expression.segments.order_by('position_in_expr')
for expr_seg, form_seg in itertools.izip(expr_segments, form_segments):
orth = form_seg
if expr_seg.has_nps:
expr += orth
else:
expr += ' %s' % orth
return expr.lstrip()
def get_normalized_expr_query(expression, form_segments):
expr = ''
expr_segments = expression.segments.order_by('position_in_expr')
for expr_seg, form_seg in itertools.izip(expr_segments, form_segments):
orth = form_seg
if expr_seg.ctag == 'interp':
orth = u'[orth="\\%s"]' % orth
else:
orth = orth + '/i'
if expr_seg.has_nps:
expr += orth
else:
expr += ' %s' % orth
return expr.lstrip()
def segment(orig_seg, synonym):
return generate_inflected_forms(synonym, orig_seg)
def generate_inflected_forms(synonym, orig_seg):
inflected = []
orig_morf2_interps = settings.MORFEUSZ2.analyse(orig_seg.orth.encode('utf8'))
for interp in orig_morf2_interps:
if interp.getTag(settings.MORFEUSZ2) == u'%s:%s' % (orig_seg.ctag, orig_seg.msd):
inflected.extend(inflect(synonym, interp.tagId))
return set(inflected)
def inflect(base, tag_id):
possible_forms = settings.MORFEUSZ2.generate(base.encode('utf8'), tag_id)
return [form.orth for form in possible_forms]