|
|
|
import numpy as np
|
|
|
|
import scipy.sparse as sp
|
|
|
|
|
|
|
|
|
|
|
|
def sparse_to_tuple(sparse_mx):
|
|
|
|
# 判断是否是coo_matrix,不是的话就转成coo_matrix
|
|
|
|
if not sp.isspmatrix_coo(sparse_mx):
|
|
|
|
sparse_mx = sparse_mx.tocoo()
|
|
|
|
coords = np.vstack((sparse_mx.row, sparse_mx.col)).transpose()
|
|
|
|
values = sparse_mx.data
|
|
|
|
shape = sparse_mx.shape
|
|
|
|
return coords, values, shape
|
|
|
|
|
|
|
|
|
|
|
|
def preprocess_graph(adj):
|
|
|
|
# A.sum(axis=1):计算矩阵的每一行元素之和,得到节点的度矩阵D
|
|
|
|
# np.power(x, n):数组元素求n次方,得到D^(-1/2)
|
|
|
|
# sp.diags()函数根据给定的对象创建对角矩阵,对角线上的元素为给定对象中的元素
|
|
|
|
adj = sp.coo_matrix(adj)
|
|
|
|
adj_ = adj + sp.eye(adj.shape[0]) # A* = A+I,即对邻接矩阵加入自连接
|
|
|
|
|
|
|
|
rowsum = np.array(adj_.sum(1)) # 对行求和,即得到节点的度
|
|
|
|
degree_mat_inv_sqrt = sp.diags(np.power(rowsum, -0.5).flatten()) # 得到D的-1/2次方矩阵d
|
|
|
|
adj_normalized = adj_.dot(degree_mat_inv_sqrt).transpose().dot(degree_mat_inv_sqrt).tocoo() # 这一步的实质是做归一化,即A* × d转置 × d
|
|
|
|
return sparse_to_tuple(adj_normalized)
|
|
|
|
|
|
|
|
def construct_feed_dict(adj_normalized, adj, features, placeholders):
|
|
|
|
# construct feed dictionary
|
|
|
|
# .update()用法就是将()内的字段增加到dict当中
|
|
|
|
feed_dict = dict() # 创建一个空字典
|
|
|
|
feed_dict.update({placeholders['features']: features})
|
|
|
|
feed_dict.update({placeholders['adj']: adj_normalized})
|
|
|
|
feed_dict.update({placeholders['adj_orig']: adj})
|
|
|
|
return feed_dict
|
|
|
|
|
|
|
|
|
|
|
|
def mask_test_edges(adj):
|
|
|
|
# Function to build test set with 10% positive links
|
|
|
|
# NOTE: Splits are randomized and results might slightly deviate from reported numbers in the paper.
|
|
|
|
# TODO: Clean up.
|
|
|
|
# sp.matrix(data,offsets)是将data的元素每列的元素,按offset里的顺序在列上进行重新排列,offset里的值是偏移量
|
|
|
|
# 具体可以参考https://blog.csdn.net/ChenglinBen/article/details/84424379
|
|
|
|
# .diagonal()就是提取对角线元素
|
|
|
|
# Remove diagonal elements删除对角线元素
|
|
|
|
adj = adj - sp.dia_matrix((adj.diagonal()[np.newaxis, :], [0]), shape=adj.shape)
|
|
|
|
# 把零元素都消除掉
|
|
|
|
adj.eliminate_zeros()
|
|
|
|
# Check that diag is zero:
|
|
|
|
# np.diag(matrix)即提取matrix的对角线元素,todense() like toarray(),区别是一个是将存储方式由稀疏矩阵转成正常矩阵,另一个是转成array
|
|
|
|
# assert检查是否对角线元素是否都被清空了
|
|
|
|
assert np.diag(adj.todense()).sum() == 0
|
|
|
|
|
|
|
|
# sp.triu(matrix)获取matrix的上三角矩阵,相应的,tril()是获取下三角矩阵
|
|
|
|
adj_triu = sp.triu(adj)
|
|
|
|
adj_tuple = sparse_to_tuple(adj_triu)
|
|
|
|
# edges相当于组合,因为是上三角矩阵的edge,所以减少了一半的重复量,(4.6)与(6,4)不会同时存在,而只会保留(4,6)
|
|
|
|
# edges_all相当于排列,就都包含了
|
|
|
|
edges = adj_tuple[0]
|
|
|
|
edges_all = sparse_to_tuple(adj)[0]
|
|
|
|
# 取edge的10%作为test
|
|
|
|
# 取edge的20%作为val
|
|
|
|
num_test = int(np.floor(edges.shape[0] / 10.))
|
|
|
|
num_val = int(np.floor(edges.shape[0] / 20.))
|
|
|
|
|
|
|
|
# 随机选取一部分作为test与val
|
|
|
|
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)
|
|
|
|
|
|
|
|
# 该函数请参考github中gae的写法,应该是更新了,这种方法应该是错的,或者说与python3不兼容
|
|
|
|
# 其中,return部分或许应该改成np.any(rows_close)
|
|
|
|
def ismember(a, b, tol=5):
|
|
|
|
# 该函数的作用就是判断a元素是否存在于b集合中
|
|
|
|
rows_close = np.all(np.round(a - b[:, None], tol) == 0, axis=-1)
|
|
|
|
return np.any(rows_close)
|
|
|
|
# return (np.all(np.any(rows_close, axis=-1), axis=-1) and
|
|
|
|
# np.all(np.any(rows_close, axis=0), axis=0))
|
|
|
|
|
|
|
|
# test_edges_false是去生成一些本来就不存在的edges
|
|
|
|
test_edges_false = []
|
|
|
|
while len(test_edges_false) < len(test_edges):
|
|
|
|
idx_i = np.random.randint(0, adj.shape[0])
|
|
|
|
idx_j = np.random.randint(0, adj.shape[0])
|
|
|
|
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
|
|
|
|
if ismember([idx_i, idx_j], np.array(test_edges_false)):
|
|
|
|
continue
|
|
|
|
test_edges_false.append([idx_i, idx_j])
|
|
|
|
|
|
|
|
# val_edges_false生成一些不存在于train与val的edges
|
|
|
|
val_edges_false = []
|
|
|
|
while len(val_edges_false) < len(val_edges):
|
|
|
|
idx_i = np.random.randint(0, adj.shape[0])
|
|
|
|
idx_j = np.random.randint(0, adj.shape[0])
|
|
|
|
if idx_i == idx_j:
|
|
|
|
continue
|
|
|
|
if ismember([idx_i, idx_j], train_edges):
|
|
|
|
continue
|
|
|
|
if ismember([idx_j, idx_i], train_edges):
|
|
|
|
continue
|
|
|
|
if ismember([idx_i, idx_j], val_edges):
|
|
|
|
continue
|
|
|
|
if ismember([idx_j, idx_i], val_edges):
|
|
|
|
continue
|
|
|
|
if val_edges_false:
|
|
|
|
if ismember([idx_j, idx_i], np.array(val_edges_false)):
|
|
|
|
continue
|
|
|
|
if ismember([idx_i, idx_j], np.array(val_edges_false)):
|
|
|
|
continue
|
|
|
|
val_edges_false.append([idx_i, idx_j])
|
|
|
|
|
|
|
|
assert ~ismember(test_edges_false, edges_all)
|
|
|
|
assert ~ismember(val_edges_false, edges_all)
|
|
|
|
assert ~ismember(val_edges, train_edges)
|
|
|
|
assert ~ismember(test_edges, train_edges)
|
|
|
|
assert ~ismember(val_edges, test_edges)
|
|
|
|
|
|
|
|
data = np.ones(train_edges.shape[0])
|
|
|
|
|
|
|
|
# Re-build adj matrix
|
|
|
|
# 如英文注释所说,这里将处理好的train_edges再重建出adj_train
|
|
|
|
adj_train = sp.csr_matrix((data, (train_edges[:, 0], train_edges[:, 1])), shape=adj.shape)
|
|
|
|
adj_train = adj_train + adj_train.T
|
|
|
|
|
|
|
|
# NOTE: these edge lists only contain single direction of edge!
|
|
|
|
return adj_train, train_edges, val_edges, val_edges_false, test_edges, test_edges_false
|