-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathprepocess-nell-v1.py
executable file
·309 lines (219 loc) · 10.7 KB
/
prepocess-nell-v1.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
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Thu May 12 15:23:18 2022
@author: pnaddaf
"""
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Thu May 12 15:09:44 2022
@author: pnaddaf
"""
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Wed May 11 15:50:17 2022
@author: pnaddaf
"""
import numpy as np
import ast
import pickle
import scipy.sparse as ss
import torch
import math
from utils import *
save_path = '/localhome/pnaddaf/Desktop/parmis/SEAl_miror/datasets_LLGF/'
dataset = 'nell_v4'
ind = ''
x_path = '/localhome/pnaddaf/Desktop/parmis/grail-master/data/' + dataset + '/entity2vec.vec'
train_path_trans = "/localhome/pnaddaf/Desktop/parmis/grail-master/data/" + dataset + "/train.txt"
test_path_trans = "/localhome/pnaddaf/Desktop/parmis/grail-master/data/" + dataset + "/test.txt"
val_path_trans = "/localhome/pnaddaf/Desktop/parmis/grail-master/data/" + dataset + "/valid.txt"
train_path_ind = "/localhome/pnaddaf/Desktop/parmis/grail-master/data/" + dataset + "_ind/train.txt"
test_path_ind = "/localhome/pnaddaf/Desktop/parmis/grail-master/data/" + dataset + "_ind/test.txt"
val_path_ind = "/localhome/pnaddaf/Desktop/parmis/grail-master/data/" + dataset + "_ind/valid.txt"
nodes_dict_path = "/localhome/pnaddaf/Desktop/parmis/grail-master/data/" + dataset + "/entity2id.txt"
def read_edges(edge_path):
edge_list = []
with open(edge_path) as f:
for line in f:
line = line.split("\t")
source = nodes_dict[line[0]]
target = nodes_dict[line[2][:-1]]
edge_list.append([source, target])
edge_list.append([target, source])
# edge_list.append([target, target])
# edge_list.append([source, source])
all_nodes.add(source)
all_nodes.add(target)
return edge_list
def vis_subgraph(x, adjacency_matri):
f = plt.figure(figsize=(20, 20))
limits = plt.axis('off')
subgraph = get_subgraph(x, adjacency_matrix)
g = torch_geometric.utils.to_networkx(subgraph[0], to_undirected=True)
nx.draw(g)
f.savefig('temp1.png')
def get_nodes_neighbourhood(adjacency_matrix, x):
adjacency_matrix = adjacency_matrix.todense()
root_list = random.sample(range(0, adjacency_matrix.shape[0]), math.floor(2/10 * adjacency_matrix.shape[0]) )
trans_neighbours = []
nodes_to_be_removed = set()
for root in root_list:
nodes_to_be_removed.add(root)
neighbour_list = adjacency_matrix[root].nonzero()[1]
for neighbour in neighbour_list:
trans_neighbours.append([root, neighbour ])
trans_neighbours.append([neighbour, root ])
nodes_to_be_removed.add(neighbour)
neighbour_of_neighbour = adjacency_matrix[neighbour].nonzero()[1]
for nn in neighbour_of_neighbour :
if nn in neighbour_list and nn != root:
trans_neighbours.append([neighbour, nn ])
adjacency_matrix[: , np.array(list(nodes_to_be_removed))] = 0
adjacency_matrix[np.array(list(nodes_to_be_removed)), :] = 0
# vis_subgraph(x, adjacency_matrix)
all_nodes = [i for i in range(adjacency_matrix.shape[0])]
trans_nodes = np.array(list(nodes_to_be_removed))
ind_nodes = []
for node in all_nodes:
if node not in trans_nodes:
ind_nodes.append(node)
ind_neighbours = []
for node in ind_nodes:
for j in adjacency_matrix[node].nonzero()[1]:
ind_neighbours.append([node, j ])
return ind_neighbours, trans_neighbours, ind_nodes, trans_nodes
def ismember(a, b, tol=5):
rows_close = np.all(np.round(a - b[:, None], tol) == 0, axis=-1)
return np.any(rows_close)
def get_false_new(test_edges, train_edges, val_edges, not_included_nodes, feat):
edge_list = np.asarray( test_edges.tolist() + train_edges.tolist() + val_edges.tolist(), dtype=np.float32)
adj = fill_adj(edge_list, feat)
adj[: , np.array(list(not_included_nodes))] = -2
adj[np.array(list(not_included_nodes)), :] = -2
false_edges = np.argwhere(adj == 0)
random.shuffle(false_edges)
test_num , train_num, val_num = len(test_edges), len(train_edges), len(val_edges)
test_edges_false = false_edges[:test_num]
train_edges_false = false_edges[test_num: test_num + train_num]
val_edges_false = false_edges[test_num + train_num: test_num + train_num + val_num]
return train_edges_false, test_edges_false, val_edges_false
def get_false_edges(test_edges, train_edges, val_edges , nodes):
edges_all = np.asarray( test_edges.tolist() + train_edges.tolist() + val_edges.tolist(), dtype=np.float32)
test_edges_false = []
while len(test_edges_false) < len(test_edges):
if len(test_edges_false) % 1000 == 0:
print( len(test_edges_false))
idx_i = random.choice(nodes)
idx_j = random.choice(nodes)
if idx_i == idx_j:
continue
if ismember([idx_i, idx_j], edges_all):
continue
if test_edges_false:
if ismember([idx_j, idx_i], np.array(test_edges_false)):
continue
test_edges_false.append([idx_i, idx_j])
print("Finished False Test")
val_edges_false = []
while len(val_edges_false) < len(val_edges):
if len(val_edges_false) % 1000 == 0:
print( len(val_edges_false))
idx_i = random.choice(nodes)
idx_j = random.choice(nodes)
if idx_i == idx_j:
continue
if ismember([idx_i, idx_j], edges_all):
continue
if ismember([idx_i, idx_j], np.array(test_edges_false)):
continue
if val_edges_false:
if ismember([idx_j, idx_i], np.array(val_edges_false)):
continue
val_edges_false.append([idx_i, idx_j])
print("Finished False Val")
train_edges_false = []
# while len(train_edges_false) < len(train_edges):
# idx_i = random.choice(nodes)
# idx_j = random.choice(nodes)
# if idx_i == idx_j:
# continue
# if ismember([idx_i, idx_j], edges_all):
# continue
# if ismember([idx_i, idx_j], np.array(val_edges_false)):
# continue
# if ismember([idx_i, idx_j], np.array(test_edges_false)):
# continue
# if train_edges_false:
# if ismember([idx_j, idx_i], np.array(train_edges_false)):
# continue
# train_edges_false.append([idx_i, idx_j])
assert ~ismember(test_edges_false, edges_all)
assert ~ismember(val_edges_false, edges_all)
return train_edges_false, test_edges_false, val_edges_false
def get_pos_edges_from_edges(edges):
edges = np.array(edges)
num_test = int(np.floor(edges.shape[0] / 5.))
num_val = int(np.floor(edges.shape[0] / 10.))
all_edge_idx = list(range(edges.shape[0]))
np.random.shuffle(all_edge_idx)
val_edge_idx = all_edge_idx[:num_val]
test_edge_idx = all_edge_idx[num_val:(num_val + num_test)]
test_edges = edges[test_edge_idx]
val_edges = edges[val_edge_idx]
train_edges = np.delete(edges, np.hstack([test_edge_idx, val_edge_idx]), axis=0)
return train_edges, val_edges, test_edges
def fill_adj(edge_list, feat_data):
edge_weight = torch.ones(edge_list.shape[0], dtype=int)
adjacency_matrix = ss.csr_matrix(
(edge_weight, (torch.tensor(edge_list[:,0]), torch.tensor(edge_list[:,1]))),
shape=(len(feat_data), len(feat_data)))
return adjacency_matrix
nodes_dict = dict()
with open(nodes_dict_path) as f:
for line in f:
line = line.split("\t")
line_0 = line[0].replace("_", ":",2)
nodes_dict[line_0] = int(line[1][:-1])
feat_data = []
with open(x_path) as f:
for line in f:
feat_data.append([float(i) for i in line.split("\t")[:-1]])
np.save(save_path + 'LLGF_' + dataset + ind + '_x.npy', np.array(feat_data))
all_nodes = set()
print("Read data")
train_edges = read_edges(train_path_trans) + read_edges(train_path_ind)
test_edges = read_edges(test_path_trans) + read_edges(test_path_ind)
val_edges = read_edges(val_path_trans) + read_edges(val_path_ind)
all_edges = np.array(train_edges + test_edges + val_edges)
print("Making adj matrix")
adjacency_matrix = fill_adj(all_edges, feat_data)
print("Extracting transductive and inductive neighbourhoods")
ind_neighbours, trans_neighbours, ind_nodes, trans_nodes = get_nodes_neighbourhood(adjacency_matrix, feat_data)
print("Getting positive edges")
transductive_train_edges, transductive_valid_edges, transductive_test_edges = get_pos_edges_from_edges(trans_neighbours)
inductive_train_edges, inductive_valid_edges, inductive_test_edges = get_pos_edges_from_edges(ind_neighbours)
# get false edges
print("Getting false edges")
transductive_train_edges_false, transductive_test_edges_false, transductive_val_edges_false = get_false_new(transductive_test_edges, transductive_train_edges, transductive_valid_edges, ind_nodes, feat_data)
inductive_train_edges_false, inductive_test_edges_false, inductive_val_edges_false = get_false_new(inductive_test_edges, inductive_train_edges, inductive_valid_edges, trans_nodes , feat_data )
# transductive_train_edges_false, transductive_test_edges_false, transductive_val_edges_false = get_false_edges(transductive_test_edges, transductive_train_edges, transductive_valid_edges, trans_nodes )
# inductive_train_edges_false, inductive_test_edges_false, inductive_val_edges_false = get_false_edges(inductive_test_edges, inductive_train_edges, inductive_valid_edges, ind_nodes )
# save files
save_path = '/localhome/pnaddaf/Desktop/parmis/SEAl_miror/datasets_LLGF/'
dataset = 'Nellv1_new'
print("Saving to files")
np.save(save_path + 'LLGF_' + dataset + '_x.npy', np.array(feat_data))
np.save(save_path + 'LLGF_' + dataset + 'ind_x.npy', np.array(feat_data))
np.save(save_path + 'LLGF_' + dataset + '_train_pos.npy', np.array(transductive_train_edges))
np.save(save_path + 'LLGF_' + dataset + '_test_pos.npy', np.array(transductive_test_edges))
np.save(save_path + 'LLGF_' + dataset + '_test_neg.npy', np.array(transductive_test_edges_false))
np.save(save_path + 'LLGF_' + dataset + '_val_pos.npy', np.array(transductive_valid_edges))
np.save(save_path + 'LLGF_' + dataset + '_val_neg.npy', np.array(transductive_val_edges_false))
np.save(save_path + 'LLGF_' + dataset + 'ind_train_pos.npy', np.array(inductive_train_edges))
np.save(save_path + 'LLGF_' + dataset + 'ind_test_pos.npy', np.array(inductive_test_edges))
np.save(save_path + 'LLGF_' + dataset + 'ind_test_neg.npy', np.array(inductive_test_edges_false))
np.save(save_path + 'LLGF_' + dataset + 'ind_val_pos.npy', np.array(inductive_valid_edges))
np.save(save_path + 'LLGF_' + dataset + 'ind_val_neg.npy', np.array(inductive_val_edges_false))