-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathQuick_Main.py
210 lines (163 loc) · 6.35 KB
/
Quick_Main.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
from __future__ import division, unicode_literals
from sklearn import metrics
from sklearn.metrics import r2_score
#from textblob import TextBlob
import numpy as np
import glob
import json
import os
from sklearn.cluster import KMeans
from sklearn.cluster import AgglomerativeClustering
import sortedcollections
from sklearn.metrics.pairwise import cosine_similarity
from itertools import combinations
import itertools
def info(eval_dir):
path=os.path.join(eval_dir,"info.json")
s=open(path,'r')
data = json.load(s)
dict_file = {}
for i in range(len(data)):
dict_file[data[i][str("folder")]] = data[i][str("language")]
return dict_file
def clustering_word2vec(data_features,clustering_type='Agglomerative',optimizer_type = 'silhouette'):
d={}
# Finding the scores for cluster number ranging from 6 to Number of documents in a problem folder
for i in range(6,len(data_features)):
if clustering_type == 'KMeans':
spectral = KMeans(n_clusters=i).fit((np.array(data_features)))
if clustering_type == 'Agglomerative':
spectral = AgglomerativeClustering(n_clusters=i, linkage='ward').fit(np.array(data_features))
label = spectral.fit_predict((np.array(data_features)))
if optimizer_type == 'silhouette':
score= metrics.silhouette_score((np.array(data_features)), label,metric='euclidean')
if optimizer_type == 'calinski':
score=metrics.calinski_harabasz_score((np.array(data_features)), label)
d[i]=score
# Finding the cluster Number with the highest score
n_c=0
for key,val in d.items():
if(val==max(d.values())):
n_c=key
break
print("_"*100)
print("Optimized Cluster Number: ", n_c)
print("_"*100)
# Finally choosing the optimized cluster Number for doing the final clustering
if clustering_type == 'KMeans':
spectral == KMeans(n_clusters=n_c).fit((np.array(data_features)))
if clustering_type == 'Agglomerative':
spectral = AgglomerativeClustering(n_clusters=n_c, linkage='ward').fit(np.array(data_features))
label = spectral.fit_predict((np.array(data_features)))
return label
def prod_output(eval_dir,out_dir,k,labels):
prblm_path=os.path.join(eval_dir,k)
doc_path=glob.glob(prblm_path + '/*')
doc_list_name=[]
for i in doc_path:
m=i.split('/')
m=m[-1]
doc_list_name.append(m)
dic={}
for i,j in zip(doc_list_name,labels):
dic[i]=j
list_all = []
list_val = []
for v in dic.values():
list_val.append(v)
set_val = set(list_val)
for val in set_val:
list_per_cluster = []
for key, value in dic.items():
if val == value:
list_per_cluster.append(key)
list_all.append(list_per_cluster)
list_all_output = []
for i in range(len(list_all)):
list_cluster = []
for j in range(len(list_all[i])):
dict_per_doc = {}
dict_per_doc["document"] = list_all[i][j]
list_cluster.append(dict_per_doc)
list_all_output.append(list_cluster)
if(os.path.exists(out_dir+"/"+k)==False):
os.mkdir(out_dir+"/"+k)
out_folder=out_dir + '/' +k
out_path = out_folder + "/clustering.json"
out_file = open(out_path, "w")
json.dump(list_all_output, out_file, indent=4)
return list_all_output
def similarity_score(list_all, dict_features):
list_all_comb = []
for i in range(len(list_all)):
list_comb_percluster = []
if len(list_all[i]) > 1:
for j in range(len(list_all[i])):
list_comb_percluster.append(list_all[i][j]["document"])
list_all_comb.append(list_comb_percluster)
combs = []
for i in range(len(list_all_comb)):
comb = list(combinations(list_all_comb[i], 2))
combs.append(comb)
comb_list = list(itertools.chain(*combs))
all_sim = []
for i in range(len(comb_list)):
doc1 = comb_list[i][0].split(",")
doc2 = comb_list[i][1].split(",")
vec1 = dict_features[doc1[0]]
vec2 = dict_features[doc2[0]]
vec1=[vec1]
vec2=[vec2]
sim = cosine_similarity(vec1, vec2)
all_sim.append(sim)
return comb_list, all_sim
def write_ranking(comb_list, all_sim, out_dir,k):
list_all_output = []
for i in range(len(comb_list)):
dict_sim_perpair = {}
dict_sim_perpair["document1"] = comb_list[i][0]
dict_sim_perpair["document2"] = comb_list[i][1]
dict_sim_perpair["score"] = round(all_sim[i][0][0],6)
list_all_output.append(dict_sim_perpair)
out_folder=out_dir + '/' +k
out_path = out_folder + "/ranking.json"
out_file = open(out_path, "w")
json.dump(list_all_output, out_file, indent=4)
return list_all_output
if __name__=="__main__":
eval_dir="./quick_run_main/pan17-author-clustering-test-dataset-2017-03-14"
out_dir="./quick_run_main/tfidf_weight_x_word2vec_output_pan17-author-clustering-test-dataset-2017-03-14"
dict_f=info(eval_dir)
dict_f=sortedcollections.OrderedDict(sorted(dict_f.items()))
en_=open("./quick_run_main/english_weighted.json",'r')
dt_=open("./quick_run_main/dutch_weighted.json",'r')
gr_=open("./quick_run_main/greek_weighted.json",'r')
vec_en=json.load(en_)
vec_dt=json.load(dt_)
vec_gr=json.load(gr_)
for k,v in dict_f.items():
if v=="en":
vectors=vec_en[k]
labels=clustering_word2vec(vectors)
elif v=="nl":
vectors=vec_dt[k]
labels=clustering_word2vec(vectors)
elif v=="gr":
vectors=vec_gr[k]
labels=clustering_word2vec(vectors)
list_all=prod_output(eval_dir,out_dir,k,labels)
dict_features={}
prblm_path=os.path.join(eval_dir,k)
doc_path=glob.glob(prblm_path + '/*')
doc_list_name=[]
for i in doc_path:
m=i.split('/')
m=m[-1]
doc_list_name.append(m)
i=0
for j in doc_list_name:
dict_features[j]=vectors[i]
i=i+1
# similarity between documents
list_comb, all_sim = similarity_score(list_all, dict_features)
list_sim = write_ranking(list_comb, all_sim, out_dir,k)