/
NLP.py
356 lines (283 loc) · 9.44 KB
/
NLP.py
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
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
'''
FRAMES
======
frame() to look up a frame by its exact name or ID
frames() to get frames matching a name pattern
frames_by_lemma() to get frames containing an LU matching a name pattern
frame_ids_and_names() to get a mapping from frame IDs to names
FRAME ELEMENTS
==============
fes() to get frame elements (a.k.a. roles) matching a name pattern, optionally constrained
by a frame name pattern
LEXICAL UNITS
=============
lu() to look up an LU by its ID
lus() to get lexical units matching a name pattern, optionally constrained by frame
lu_ids_and_names() to get a mapping from LU IDs to names
'''
from nltk.corpus import framenet as fn
import collections
import numpy as np
import math
import networkx as nx
import matplotlib.pyplot as plt
result1 = []
result2 = []
simres1 = []
simres2 = []
def getFrame_by_Name(frame_name):
f = fn.frame_by_name(frame_name)
pass
def getFE(frame_name):
f = fn.frame_by_name(frame_name)
FE = f.FE
pass
def fekeys(frame_name):
f = fn.frame_by_name(frame_name)
FE = f.FE
keys = FE.keys()
pass
def items():
f = fn.frame_by_name(frame_name)
FE = f.FE
items = FE.items()
pass
def fe_coreTypes(frame_name):
f = fn.frame_by_name(frame_name)
fedict = {}
for feName, fe in f.FE.items():
fedict[feName] = fe.coreType
print(feName)
print(fe)
return fedict
def exemplars1(frame_name, feR=False):
expls = []
for i in lu_ids(frame_name):
e = fn.lu(i).exemplars
for j in range(len(e)):
#print('+++++++', e[j].FE)
expls.append(e[j])
pass
pass
annotation1(expls,frame_name)
return simres1
pass
def exemplars2(frame_name, feR=False):
expls = []
for i in lu_ids(frame_name):
e = fn.lu(i).exemplars
for j in range(len(e)):
#print('+++++++', e[j].FE)
expls.append(e[j])
pass
pass
annotation2(expls,frame_name)
return simres2
pass
# get all lu id of a frame
def lu_ids(frame_name):
f = fn.frame_by_name(frame_name)
ids = []
for v in f.lexUnit.values():
ids.append(v.ID)
return ids
def annotation1(exemplars,frame_name):
num = 0
for sentence in exemplars:
num = num+1
_exemplar_of_POS1(sentence,False)
pass
#res = set()
for feName1 in result1:
count = 0
for feName2 in result1:
if feName1.__getitem__('FEName') == feName2.__getitem__('FEName') and feName1.__getitem__('POS') == feName2.__getitem__('POS'):
count=count+1
pass
pass
#res.add(str(feName1.__getitem__('FEName'))+" "+str(feName1.__getitem__('POS'))+" "+str(count))
pair = {"FEName" : feName1.__getitem__('FEName'), "POS" : feName1.__getitem__('POS'), "Count" : count}
if pair not in simres1:
simres1.append(pair)
pass
pass
'''
file = open('report.txt', 'a')
file.write(str(frame_name) + " " + "Total sentences: "+str(num))
file.write("\n")
file.write(str(res))
file.write("\n")
file.write("\n")
file.close()
'''
pass
def annotation2(exemplars,frame_name):
num = 0
for sentence in exemplars:
num = num+1
_exemplar_of_POS2(sentence,False)
pass
#res = set()
for feName1 in result2:
count = 0
for feName2 in result2:
if feName1.__getitem__('FEName') == feName2.__getitem__('FEName') and feName1.__getitem__('POS') == feName2.__getitem__('POS'):
count=count+1
pass
pass
#res.add(str(feName1.__getitem__('FEName'))+" "+str(feName1.__getitem__('POS'))+" "+str(count))
pair = {"FEName" : feName1.__getitem__('FEName'),"POS" : feName1.__getitem__('POS'), "Count" : count}
if pair not in simres2:
simres2.append(pair)
pass
pass
'''
file = open('report.txt', 'a')
file.write(str(frame_name) + " " + "Total sentences: "+str(num))
file.write("\n")
file.write(str(res))
file.write("\n")
file.write("\n")
file.close()
'''
pass
'''
def aggregate_names(errors,pair):
pair = {"FEName" : feName1.__getitem__('FEName'), "POS" : feName1.__getitem__('POS'), "Count" : count}
result = collections.defaultdict(lambda: collections.defaultdict(pair))
for real_name, false_name, location in errors:
result[real_name][false_name].append(location)
return result
'''
def _exemplar_of_POS1(sentence, fe_only = True):
#fe = sentence.FE[0]
#print(list(zip(*sentence.FE[0]))[2] if sentence.FE[0] else set()) # [(0, 68, 'Responsible_party'), (84, 89, 'Responsible_party'), (90, 95, 'Station')]
#num = len(fe)
#print(">>>>>>flagged: ", sentence.PT) #sentence.POS
overtNames = list(zip(*sentence.FE[0]))[2] if sentence.FE[0] else "" # {'Station', 'Responsible_party'}
Pos_Tag = list(zip(*sentence.PT))[2] if sentence.PT else "" # {'NP', 'Poss'}
Gram_Tag = list(zip(*sentence.GF))[2] if sentence.GF else "" # {'Ext', 'Gen'}
minlength = min(len(overtNames),len(Pos_Tag),len(Gram_Tag))
for i in range(minlength):
#print(i) #overtNames[0] = Responsible_party, overtNames[1] = Responsible_party, overtNames[2] = station
try:
pair = {"FEName" : overtNames[i], "POS" : str(Pos_Tag[i])+"."+str(Gram_Tag[i])}
result1.append(pair)
except IndexError:
pass
pass
pass
def _exemplar_of_POS2(sentence, fe_only = True):
#fe = sentence.FE[0]
#print(list(zip(*sentence.FE[0]))[2] if sentence.FE[0] else set()) # [(0, 68, 'Responsible_party'), (84, 89, 'Responsible_party'), (90, 95, 'Station')]
#num = len(fe)
#print(">>>>>>flagged: ", sentence.PT) #sentence.POS
overtNames = list(zip(*sentence.FE[0]))[2] if sentence.FE[0] else "" # {'Station', 'Responsible_party'}
Pos_Tag = list(zip(*sentence.PT))[2] if sentence.PT else "" # {'NP', 'Poss'}
Gram_Tag = list(zip(*sentence.GF))[2] if sentence.GF else "" # {'Ext', 'Gen'}
minlength = min(len(overtNames),len(Pos_Tag),len(Gram_Tag))
for i in range(minlength):
#print(i) #overtNames[0] = Responsible_party, overtNames[1] = Responsible_party, overtNames[2] = station
try:
pair = {"FEName" : overtNames[i], "POS" : str(Pos_Tag[i])+"."+str(Gram_Tag[i])}
result2.append(pair)
except IndexError:
pass
pass
pass
def bipartie_match(frame_name1,frame_name2):
G = nx.Graph()#
for feName1 in simres1:
for feName2 in simres2:
str1 = "F1" + " "+feName1.__getitem__('FEName')
str2 = "F2" + " "+feName2.__getitem__('FEName')
edge_weight = calc_sim(feName1,feName2)
G.add_edge(str1,str2,weight = edge_weight)
pass
pass
res=nx.max_weight_matching(G, maxcardinality=True, weight='weight')
print(str(frame_name1) + " " + str(frame_name2))
print(res)
total_weight=0
for edges in res:
total_weight = total_weight + (G[edges[0]][edges[1]]['weight']) #get edge weight value
pass
print(total_weight)
file = open('Frame_Similarity.txt', 'a')
file.write(str(frame_name1) + " " + str(frame_name2))
file.write("\n")
file.write(str(total_weight))
file.write("\n")
file.write(str(res))
file.write("\n")
file.write("\n")
file.write("\n")
file.close()
pass
def calc_sim(feName1,feName2):
sim_value = 0
#case1 name match, pos match, then calc the sim corelated to the Count
if feName1.__getitem__('FEName') == feName2.__getitem__('FEName') and feName1.__getitem__('POS') == feName2.__getitem__('POS'):
sim_value = 2*(prob_distribution1(feName1))*(prob_distribution2(feName2))
#sim_value = 2*(feName1.__getitem__('Count')+feName2.__getitem__('Count'))
return sim_value
pass
#case2 name match, pos not match
if feName1.__getitem__('FEName') == feName2.__getitem__('FEName') and feName1.__getitem__('POS') != feName2.__getitem__('POS'):
sim_value = 1*(prob_distribution1(feName1))*(prob_distribution2(feName2))
#sim_value = 1*(feName1.__getitem__('Count')+feName2.__getitem__('Count'))
return sim_value
pass
#case3 name not match, pos match
if feName1.__getitem__('FEName') != feName2.__getitem__('FEName') and feName1.__getitem__('POS') == feName2.__getitem__('POS'):
sim_value = 0.5*(prob_distribution1(feName1))*(prob_distribution2(feName2))
#sim_value = 0.5*(feName1.__getitem__('Count')+feName2.__getitem__('Count'))
return sim_value
pass
#case4 name not match, pos not match
if feName1.__getitem__('FEName') != feName2.__getitem__('FEName') and feName1.__getitem__('POS') != feName2.__getitem__('POS'):
sim_value = 0
return sim_value
pass
return sim_value
pass
def cosine_sim(fileid1,fileid2):#calc cosine simlirity
vec1 = fileid1.__getitem__('Pos_Dist') #get tf-idf vec for fileid1
vec2 = fileid2.__getitem__('Pos_Dist') #get tf-idf vec for fileid1
norm1 = normalizer(vec1) #normalize vec1
norm2 = normalizer(vec2) #normalize vec2
sim = np.dot(vec1, vec2) / (norm1*norm2) #calc cosine simlirity
return sim
pass
def normalizer(self,vec): # vector normalize
denom = np.sum([el**2 for el in vec])
return math.sqrt(denom)
pass
def prob_distribution1(FEName_Target):
target_count = 0
total_count = 0
for feName1 in simres1:
if feName1.__getitem__('FEName') == FEName_Target.__getitem__('FEName'):
total_count = total_count + feName1.__getitem__('Count')
pass
if feName1.__getitem__('FEName') == FEName_Target.__getitem__('FEName') and feName1.__getitem__('POS') == FEName_Target.__getitem__('POS'):
target_count = feName1.__getitem__('Count')
pass
pass
prob = target_count/total_count
return prob
pass
def prob_distribution2(FEName_Target):
target_count = 0
total_count = 0
for feName1 in simres2:
if feName1.__getitem__('FEName') == FEName_Target.__getitem__('FEName'):
total_count = total_count + feName1.__getitem__('Count')
pass
if feName1.__getitem__('FEName') == FEName_Target.__getitem__('FEName') and feName1.__getitem__('POS') == FEName_Target.__getitem__('POS'):
target_count = feName1.__getitem__('Count')
pass
pass
prob = target_count/total_count
return prob
pass