forked from danielpreotiuc/textrank
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtextrank.py
executable file
·252 lines (226 loc) · 8.68 KB
/
textrank.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
import sys
import langid
import os
import nltk
from nltk.tag.api import TaggerI
from nltk.internals import find_file, find_jar, config_java, java, _java_options, find_jars_within_path
import itertools
from operator import itemgetter
from nltk.stem import WordNetLemmatizer
import networkx as nx
from nltk.collocations import *
from nltk.stem.porter import *
tagger = nltk.tag.perceptron.PerceptronTagger()
wnl = WordNetLemmatizer()
colloc_list = []
entity_names = []
def filter_for_tags(tagged, tags=['NN', 'NNPS', 'NNP', 'NNS']):
return [item for item in tagged if item[1] in tags]
def filter_numbers(tagged):
return [item for item in tagged if len(item[0])>2 and not item[0].isdigit()]
def normalize(tagged):
return [(item[0].replace('.', ''), item[1]) for item in tagged]
def normalize_tags(tagged):
return [(item[0], item[1][0:1]) for item in tagged]
def lowercase(tagged):
return [(w.lower(),t) for (w,t) in tagged]
def rstopwords(tagged):
return [(w,t) for (w,t) in tagged if not w in nltk.corpus.stopwords.words('english')]
def lemmatize(tagged):
return [(wnl.lemmatize(item[0]),item[1]) if not ' ' in item[0] else (item[0],item[1]) for item in tagged]
def extract_entity_names(t):
entity_names = []
if hasattr(t, 'label') and t.label:
if t.label() == 'NE':
entity_names.append(' '.join([child[0] for child in t]))
else:
for child in t:
entity_names.extend(extract_entity_names(child))
return entity_names
def joincolloc(tagged):
tagged1 = []
sw = 0
for i in range(len(tagged)-1):
if sw == 1:
sw = 0
continue
if (tagged[i],tagged[i+1]) in colloc_list:
sw = 1
if tagged[i][1].startswith('NN') or tagged[i+1][1].startswith('NN'):
tagged1.append((tagged[i][0]+' '+tagged[i+1][0],'NN'))
elif tagged[i][1]=='RB' or tagged[i+1][1]=='RB':
tagged1.append((tagged[i][0]+' '+tagged[i+1][0],'RB'))
else:
tagged1.append((tagged[i][0]+' '+tagged[i+1][0],tagged[i][1]))
else:
tagged1.append(tagged[i])
if len(tagged)>0:
tagged1.append(tagged[len(tagged)-1])
return tagged1
def groupne2(tagged):
tagged1 = []
sw = 0
for i in range(len(tagged)-1):
if sw == 1:
sw = 0
continue
if (tagged[i][0]+' '+tagged[i+1][0]) in entity_names:
sw = 1
if tagged[i][1]=='NNP' or tagged[i+1][1]=='NNP':
tagged1.append((tagged[i][0]+' '+tagged[i+1][0],'NNP'))
elif tagged[i][1]=='NN' or tagged[i+1][1]=='NN':
tagged1.append((tagged[i][0]+' '+tagged[i+1][0],'NN'))
elif tagged[i][1]=='RB' or tagged[i+1][1]=='RB':
tagged1.append((tagged[i][0]+' '+tagged[i+1][0],'RB'))
else:
tagged1.append((tagged[i][0]+' '+tagged[i+1][0],tagged[i][1]))
else:
tagged1.append(tagged[i])
if len(tagged)>0:
tagged1.append(tagged[len(tagged)-1])
return tagged1
def groupne3(tagged):
tagged1 = []
sw = 0
for i in range(len(tagged)-2):
if sw == 1:
sw = 0
continue
if (tagged[i][0]+' '+tagged[i+1][0]+' '+tagged[i+2][0]) in entity_names:
sw = 1
if tagged[i][1]=='NNP' or tagged[i+1][1]=='NNP' or tagged[i+2][1]=='NNP':
tagged1.append((tagged[i][0]+' '+tagged[i+1][0]+' '+tagged[i+2][0],'NNP'))
elif tagged[i][1]=='NN' or tagged[i+1][1]=='NN' or tagged[i+2][1]=='NNP':
tagged1.append((tagged[i][0]+' '+tagged[i+1][0]+' '+tagged[i+2][0],'NN'))
elif tagged[i][1]=='RB' or tagged[i+1][1]=='RB' or tagged[i+2][1]=='NNP':
tagged1.append((tagged[i][0]+' '+tagged[i+1][0]+' '+tagged[i+2][0],'RB'))
else:
tagged1.append((tagged[i][0]+' '+tagged[i+1][0]+' '+tagged[i+2][0],tagged[i][1]))
else:
tagged1.append(tagged[i])
if len(tagged)>1:
tagged1.append(tagged[len(tagged)-2])
tagged1.append(tagged[len(tagged)-1])
elif len(tagged)==1:
tagged1.append(tagged[len(tagged)-1])
return tagged1
def joincollocbi(tagged):
tagged1 = []
sw = 0
for i in range(len(tagged)-1):
if sw == 1:
sw = 0
continue
if ' ' in tagged[i][0]:
t1 = (tagged[i][0][tagged[i][0].find(' '):].strip(), tagged[i][1])
else:
t1 = (tagged[i][0], tagged[i][1])
if ' ' in tagged[i+1][0]:
t2 = (tagged[i+1][0][:tagged[i+1][0].find(' ')].strip(), tagged[i][1])
else:
t2 = (tagged[i+1][0], tagged[i+1][1])
if (t1,t2) in colloc_list:
sw = 1
if tagged[i][1]=='NNP' or tagged[i+1][1]=='NNP':
tagged1.append((tagged[i][0]+' '+tagged[i+1][0],'NNP'))
elif tagged[i][1]=='NN' or tagged[i+1][1]=='NN':
tagged1.append((tagged[i][0]+' '+tagged[i+1][0],'NN'))
elif tagged[i][1]=='RB' or tagged[i+1][1]=='RB':
tagged1.append((tagged[i][0]+' '+tagged[i+1][0],'RB'))
else:
tagged1.append((tagged[i][0]+' '+tagged[i+1][0],tagged[i][1]))
else:
tagged1.append(tagged[i])
if len(tagged)>0:
tagged1.append(tagged[len(tagged)-1])
return tagged1
blacklist = []
fname=sys.argv[1]
articles = os.listdir(fname)
FOLDER = 'keywords-'+fname+'-textrank'
if not os.path.exists(FOLDER): os.makedirs(FOLDER)
tagged = []
for article in articles:
articleFile = open(fname+'/' + article, 'r')
for linee in articleFile:
line=linee.decode('latin-1')
lang = langid.classify(line.strip())
if not lang[0]=='en':
continue
sentences = nltk.sent_tokenize(line.strip())
tokenized_sentences = [nltk.word_tokenize(sentence) for sentence in sentences]
tagged_sentences = [tagger.tag(sentence) for sentence in tokenized_sentences]
for sentence in tagged_sentences:
tagged.extend(sentence)
chunked_sentences = nltk.ne_chunk_sents(tagged_sentences, binary=True)
for tree in chunked_sentences:
entity_names.extend(extract_entity_names(tree))
articleFile.close()
entity_names = set(entity_names)
bigram_measures = nltk.collocations.BigramAssocMeasures()
finder = nltk.collocations.BigramCollocationFinder.from_words(tagged)
finder.apply_freq_filter(3)
colloc_list = finder.nbest(bigram_measures.pmi, 20) # this needs to be tweaked
for article in articles:
print 'Reading articles/' + article
articleFile = open(fname + '/' + article, 'r')
tagged=[]
sentences=[]
k=0
for linee in articleFile:
line = linee.decode('latin-1')
lang = langid.classify(line.strip())
if not lang[0]=='en':
continue
sents = nltk.sent_tokenize(line.strip())
tok_sents = [nltk.word_tokenize(sent) for sent in sents]
tagged_sents = [tagger.tag(sent) for sent in tok_sents]
tagged_sents = [joincolloc(sent) for sent in tagged_sents]
tagged_sents = [joincollocbi(sent) for sent in tagged_sents]
tagged_sents = [groupne2(sent) for sent in tagged_sents]
tagged_sents = [groupne3(sent) for sent in tagged_sents]
tagged_sents = [filter_for_tags(sent) for sent in tagged_sents]
tagged_sents = [normalize_tags(sent) for sent in tagged_sents]
tagged_sents = [normalize(sent) for sent in tagged_sents]
tagged_sents = [filter_numbers(sent) for sent in tagged_sents]
tagged_sents = [lowercase(sent) for sent in tagged_sents]
tagged_sents = [lemmatize(sent) for sent in tagged_sents]
tagged_sents = [rstopwords(sent) for sent in tagged_sents]
for sent in tagged_sents:
tagged.extend(sent)
sentences.extend(tagged_sents)
gr = nx.MultiGraph()
for sentence in sentences:
if len(sentence)>1:
for i in range(len(sentence)-1):
for j in range(i+1,len(sentence)):
try:
s1 = sentence[i][0] + '/' + sentence[i][1]
s2 = sentence[j][0] + '/' + sentence[j][1]
# wt = float(1.0)/float(len(sentence)) # if weighting by sentence length is desired
wt = 1
gr.add_edge(s1,s2,weight=wt)
except AdditionError, e:
pass
H=nx.Graph()
for u,v,d in gr.edges(data=True):
w = d['weight']
if H.has_edge(u,v):
H[u][v]['weight'] += w
else:
H.add_edge(u,v,weight=w)
calculated_page_rank = nx.pagerank(H)
keyphraseFile = open(FOLDER + '/'+article, 'w')
di = sorted(calculated_page_rank.iteritems(), key=itemgetter(1), reverse=True)
dic = []
for k, g in itertools.groupby(di, key=itemgetter(1)):
try:
w = str(map(itemgetter(0), g)[0])
w = w[:w.find('/')]
if len(w)>2 and w not in blacklist:
if w not in dic:
keyphraseFile.write(w.replace(' ','_') + ':' + str(k)[0:6] + '\n')
dic.append(w)
except:
pass
keyphraseFile.close()