/
ck12_wiki_predict.py
163 lines (124 loc) · 5.21 KB
/
ck12_wiki_predict.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
import argparse
import utils
import numpy as np
from scipy import linalg
import pickle
import pandas as pd
from nltk.corpus import stopwords
import gensim, logging, os
path = '/home/evan/Desktop/Kaggle/allen/glove/kaggle_allen/data/wiki_data'
stop = stopwords.words('english')
class MySentences(object):
def __init__(self,dirname):
self.dirname = dirname
def __iter__(self):
for fname in os.listdir(self.dirname):
for line in open(os.path.join(self.dirname, fname)):
yield utils.tokenize(line)
sentences = MySentences(path)
N = 1000
#model = gensim.models.Word2Vec(sentences, workers=3, size=N)
#model.save('concepts_and_wiki_data_%s' % N)
model = gensim.models.Word2Vec.load('concepts_and_wiki_data_%s' % N)
#print model['physics']
#print 'saved model'
#urls to get toppics
"""'https://www.ck12.org/earth-science/', 'http://www.ck12.org/life-science/',
'http://www.ck12.org/physical-science/', 'http://www.ck12.org/biology/',
'http://www.ck12.org/chemistry/', """
ck12_url_topic = ['http://www.ck12.org/physics/'] #
wiki_docs_dir = 'data/wiki_data'
def get_wiki_docs():
# get keywords
"""
for url_topic in ck12_url_topic:
ck12_keywords = set()
seg_dir = wiki_docs_dir + '/' + url_topic.split('/')[3]
print seg_dir
keywords= utils.get_keyword_from_url_topic(url_topic)
for kw in keywords:
ck12_keywords.add(kw)
#get and save wiki docs
utils.get_save_wiki_docs(ck12_keywords, save_folder=seg_dir)"""
ck12_keywords = set()
with open("/home/evan/Desktop/wiki_kw.txt", 'r') as f:
for line in f:
ck12_keywords.add(line.rstrip())
utils.get_save_wiki_docs(ck12_keywords, save_folder=wiki_docs_dir)
def predict(data, docs_per_q):
#index docs
docs_tf, words_idf = utils.get_docstf_idf(wiki_docs_dir)
#docs_tf = pickle.load(open('docs_tf_data.p', 'rb'))
#words_idf = pickle.load(open('words_idf_data.p', 'rb'))
pickle.dump(docs_tf, open("docs_tf_data.p", 'wb'))
pickle.dump(words_idf, open("words_idf_data.p", 'wb'))
res = []
print 'predict'
for index, row in data.iterrows():
#get answers words
w_A = set(utils.tokenize(row['answerA']))
w_B = set(utils.tokenize(row['answerB']))
w_C = set(utils.tokenize(row['answerC']))
w_D = set(utils.tokenize(row['answerD']))
A_vec = np.zeros(N)
B_vec = np.zeros(N)
C_vec = np.zeros(N)
D_vec = np.zeros(N)
sc_A = 0
sc_B = 0
sc_C = 0
sc_D = 0
print index
q = row['question']
q_vec = np.zeros(N)
for w in utils.tokenize(q):
if w in model.vocab and w not in stop:
q_vec += model[w]
q_vec = q_vec / linalg.norm(q_vec)
for d in zip(*utils.get_docs_importance_for_question(q, docs_tf, words_idf, docs_per_q))[0]:
for w in w_A:
if w in docs_tf[d]:
sc_A += 1. * docs_tf[d][w] * words_idf[w]
if w in model.vocab:
A_vec += model[w]# docs_tf (arr of tf for each doc for each word) [d] for the specific word
for w in w_B:
if w in docs_tf[d]:
sc_B += 1. * docs_tf[d][w] * words_idf[w]
if w in model.vocab:
B_vec += model[w]
for w in w_C:
if w in docs_tf[d]:
sc_C += 1. * docs_tf[d][w] * words_idf[w]
if w in model.vocab:
C_vec += model[w]
for w in w_D:
if w in docs_tf[d]:
sc_D += 1. * docs_tf[d][w] * words_idf[w]
if w in model.vocab:
D_vec = model[w]
A_vec = A_vec / linalg.norm(A_vec)
B_vec = B_vec / linalg.norm(B_vec)
C_vec = C_vec / linalg.norm(C_vec)
D_vec = D_vec / linalg.norm(D_vec)
semantic_scores = np.concatenate((A_vec, B_vec, C_vec, D_vec)).reshape(4, N).dot(q_vec)
semantic_scores[np.isnan(semantic_scores)] = 0
#print semantic_scores
combined_scores = [sc_A, sc_B, sc_C, sc_D] + semantic_scores
#print combined_scores
res.append(['A','B','C','D'][np.argmax(combined_scores)])
return res
if __name__ == '__main__':
#parsing input arguments
parser = argparse.ArgumentParser()
parser.add_argument('--fname', type=str, default='validation_set.tsv', help='file name with data')
parser.add_argument('--docs_per_q', type=int, default=20, help='number of docs to consider when ranking quesitons')
parser.add_argument('--get_data', type=int, default= 0, help='flag to get wiki data for IR')
args = parser.parse_args()
if args.get_data:
get_wiki_docs()
#read data
data = pd.read_csv('data/' + args.fname, sep = '\t' )
#predict
res = predict(data, args.docs_per_q)
#save result
pd.DataFrame({'id': list(data['id']), 'correctAnswer': res})[['id', 'correctAnswer']].to_csv("prediction_wiki_data_ss.csv", index = False)