-
Notifications
You must be signed in to change notification settings - Fork 25
/
data_utils.py
executable file
·190 lines (166 loc) · 7.15 KB
/
data_utils.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
# -*- coding: utf-8 -*-
import os
import pickle
import numpy as np
def load_word_vec(path, word2idx=None):
fin = open(path, 'r', encoding='utf-8', newline='\n', errors='ignore')
word_vec = {}
for line in fin:
tokens = line.rstrip().split()
if word2idx is None or tokens[0] in word2idx.keys():
try:
word_vec[tokens[0]] = np.asarray(tokens[1:], dtype='float32')
except:
print('WARNING: corrupted word vector of {} when being loaded from GloVe.'.format(tokens[0]))
return word_vec
def build_embedding_matrix(word2idx, embed_dim, type):
embedding_matrix_file_name = '{0}_{1}_embedding_matrix.pkl'.format(str(embed_dim), type)
if os.path.exists(embedding_matrix_file_name):
print('loading embedding_matrix:', embedding_matrix_file_name)
embedding_matrix = pickle.load(open(embedding_matrix_file_name, 'rb'))
else:
print('loading word vectors ...')
embedding_matrix = np.zeros((len(word2idx), embed_dim)) # idx 0 and 1 are all-zeros
embedding_matrix[1, :] = np.random.uniform(-1/np.sqrt(embed_dim), 1/np.sqrt(embed_dim), (1, embed_dim))
fname = '../glove.42B.300d.txt'
word_vec = load_word_vec(fname, word2idx=word2idx)
print('building embedding_matrix:', embedding_matrix_file_name)
for word, i in word2idx.items():
vec = word_vec.get(word)
if vec is not None:
# words not found in embedding index will be all-zeros.
embedding_matrix[i] = vec
pickle.dump(embedding_matrix, open(embedding_matrix_file_name, 'wb'))
return embedding_matrix
class Tokenizer(object):
def __init__(self, word2idx=None):
if word2idx is None:
self.word2idx = {}
self.idx2word = {}
self.idx = 0
self.word2idx['<pad>'] = self.idx
self.idx2word[self.idx] = '<pad>'
self.idx += 1
self.word2idx['<unk>'] = self.idx
self.idx2word[self.idx] = '<unk>'
self.idx += 1
else:
self.word2idx = word2idx
self.idx2word = {v:k for k,v in word2idx.items()}
def fit_on_text(self, text):
text = text.lower()
words = text.split()
for word in words:
if word not in self.word2idx:
self.word2idx[word] = self.idx
self.idx2word[self.idx] = word
self.idx += 1
def text_to_sequence(self, text):
text = text.lower()
words = text.split()
unknownidx = 1
sequence = [self.word2idx[w] if w in self.word2idx else unknownidx for w in words]
if len(sequence) == 0:
sequence = [0]
return sequence
class ABSADataset(object):
def __init__(self, data):
self.data = data
def __getitem__(self, index):
return self.data[index]
def __len__(self):
return len(self.data)
class ABSADatesetReader:
@staticmethod
def __read_text__(fnames):
text = ''
for fname in fnames:
fin = open(fname, 'r', encoding='utf-8', newline='\n', errors='ignore')
lines = fin.readlines()
fin.close()
for i in range(0, len(lines), 3):
text_left, _, text_right = [s.lower().strip() for s in lines[i].partition("$T$")]
aspect = lines[i + 1].lower().strip()
text_raw = text_left + " " + aspect + " " + text_right
text += text_raw + " "
return text
@staticmethod
def __read_data__(fname, tokenizer):
fin = open(fname, 'r', encoding='utf-8', newline='\n', errors='ignore')
lines = fin.readlines()
fin.close()
fin = open(fname+'.graph', 'rb')
idx2gragh = pickle.load(fin)
fin.close()
fin = open(fname+'.sentic', 'rb')
idx2gragh_s = pickle.load(fin)
fin.close()
fin = open(fname+'.graph_sdat', 'rb')
idx2gragh_sdat = pickle.load(fin)
fin.close()
all_data = []
for i in range(0, len(lines), 3):
text_left, _, text_right = [s.lower().strip() for s in lines[i].partition("$T$")]
aspect = lines[i + 1].lower().strip()
polarity = lines[i + 2].strip()
context = text_left + " " + aspect + " " + text_right
text_indices = tokenizer.text_to_sequence(text_left + " " + aspect + " " + text_right)
context_indices = tokenizer.text_to_sequence(text_left + " " + text_right)
aspect_indices = tokenizer.text_to_sequence(aspect)
left_indices = tokenizer.text_to_sequence(text_left)
polarity = int(polarity)+1
dependency_graph = idx2gragh[i]
sentic_graph = idx2gragh_s[i]
sdat_graph = idx2gragh_sdat[i]
data = {
'context': context,
'aspect': aspect,
'text_indices': text_indices,
'context_indices': context_indices,
'aspect_indices': aspect_indices,
'left_indices': left_indices,
'polarity': polarity,
'dependency_graph': dependency_graph,
'sentic_graph': sentic_graph,
'sdat_graph': sdat_graph,
}
all_data.append(data)
return all_data
def __init__(self, dataset='twitter', embed_dim=300):
print("preparing {0} dataset ...".format(dataset))
fname = {
'twitter': {
'train': './datasets/acl-14-short-data/train.raw',
'test': './datasets/acl-14-short-data/test.raw'
},
'rest14': {
'train': './datasets/semeval14/restaurant_train.raw',
'test': './datasets/semeval14/restaurant_test.raw'
},
'lap14': {
'train': './datasets/semeval14/laptop_train.raw',
'test': './datasets/semeval14/laptop_test.raw'
},
'rest15': {
'train': './datasets/semeval15/restaurant_train.raw',
'test': './datasets/semeval15/restaurant_test.raw'
},
'rest16': {
'train': './datasets/semeval16/restaurant_train.raw',
'test': './datasets/semeval16/restaurant_test.raw'
},
}
text = ABSADatesetReader.__read_text__([fname[dataset]['train'], fname[dataset]['test']])
if os.path.exists(dataset+'_word2idx.pkl'):
print("loading {0} tokenizer...".format(dataset))
with open(dataset+'_word2idx.pkl', 'rb') as f:
word2idx = pickle.load(f)
tokenizer = Tokenizer(word2idx=word2idx)
else:
tokenizer = Tokenizer()
tokenizer.fit_on_text(text)
with open(dataset+'_word2idx.pkl', 'wb') as f:
pickle.dump(tokenizer.word2idx, f)
self.embedding_matrix = build_embedding_matrix(tokenizer.word2idx, embed_dim, dataset)
self.train_data = ABSADataset(ABSADatesetReader.__read_data__(fname[dataset]['train'], tokenizer))
self.test_data = ABSADataset(ABSADatesetReader.__read_data__(fname[dataset]['test'], tokenizer))