def split_graph(G, output_dir, radio=0.8):
t_dir = output_dir
Gs = G
file_path = os.path.join(t_dir, 'graph.edgelist')
file_test_path = os.path.join(t_dir, 'graph_test.edgelist')
label_path = os.path.join(t_dir, 'label.txt')
G_train = nx.Graph()
G_test = nx.Graph()
edges = np.random.permutation(list(Gs.edges()))
nodes = set()
for a, b in edges:
if a not in nodes or b not in nodes:
G_train.add_edge(a, b)
nodes.add(a)
nodes.add(b)
else:
G_test.add_edge(a, b)
print(len(nodes), Gs.number_of_nodes())
assert len(nodes) == Gs.number_of_nodes()
assert len(nodes) == G_train.number_of_nodes()
num_test_edges = int((1 - radio) * Gs.number_of_edges())
now_number = G_test.number_of_edges()
if num_test_edges < now_number:
test_edges = list(G_test.edges())
G_train.add_edges_from(test_edges[:now_number - num_test_edges])
G_test.remove_edges_from(test_edges[:now_number - num_test_edges])
print("sample graph,origin: {} {}, train: {} {}, test: {} {}".format(
Gs.number_of_nodes(), Gs.number_of_edges(), G_train.number_of_nodes(),
G_train.number_of_edges(), G_test.number_of_nodes(),
G_test.number_of_edges()))
with utils.write_with_create(file_path) as f:
for i, j in G_train.edges():
print(i, j, file=f)
with utils.write_with_create(file_test_path) as f:
for i, j in G_test.edges():
print(i, j, file=f)
def get_wne(dataset_name, sampled_dir='', cache=True):
dataset_filename = os.path.abspath(
os.path.join('data/{}'.format(dataset_name), sampled_dir,
'graph.edgelist'))
labels = os.path.abspath(
os.path.join(os.path.dirname(dataset_filename), 'label.txt'))
save_path = os.path.abspath(
os.path.join('embeddings/{}'.format(dataset_name), sampled_dir,
'wme.embeddings'))
if (not cache) or (not os.path.exists(save_path)) or (
os.path.getmtime(save_path) < os.path.getmtime(dataset_filename)):
G = utils.load_graph(dataset_filename, label_name=None)
do_full = (G.number_of_nodes() < 10000)
eigenvalues = 'full' if do_full else 'auto'
wne = netlsd.heat(G,
timescales=np.logspace(-2, 2, 10),
eigenvalues=eigenvalues)
with utils.write_with_create(save_path) as f:
print(" ".join(map(str, wne)), file=f)
return np.loadtxt(save_path)
def sample_graph(G,
output_dir,
s_n,
times=10,
with_test=False,
radio=0.8,
feature_path=None):
if s_n is None:
s_n = int(np.sqrt(G.number_of_nodes()))
for t in range(times):
t_dir = os.path.join(output_dir, 's{}'.format(t))
n = random.randint(int(s_n / 2), 2 * s_n)
Gs = utils.random_walk_induced_graph_sampling(G, n)
mapping = dict(zip(Gs.nodes(), range(Gs.number_of_nodes())))
if feature_path is not None:
feats = sparse.load_npz(feature_path)
row = []
col = []
data = []
fr, fc = feats.nonzero()
for i, j in zip(fr, fc):
if i in mapping:
row.append(mapping[i])
col.append(j)
data.append(feats[i, j])
feats = sparse.csr_matrix((data, (row, col)),
shape=(len(mapping), feats.shape[1]))
Gs = nx.relabel_nodes(Gs, mapping)
file_path = os.path.join(t_dir, 'graph.edgelist')
file_test_path = os.path.join(t_dir, 'graph_test.edgelist')
label_path = os.path.join(t_dir, 'label.txt')
feature_save_path = os.path.join(t_dir, 'features.npz')
if feature_path is not None:
utils.write_with_create(feature_save_path)
sparse.save_npz(feature_save_path, feats)
if not with_test:
print("sample graph, nodes: {}, edges: {}, save into {}".format(
Gs.number_of_nodes(), Gs.number_of_edges(), t_dir))
with utils.write_with_create(file_path) as f:
for i, j in Gs.edges():
print(i, j, file=f)
with utils.write_with_create(label_path) as f:
for i, data in Gs.nodes(data=True):
if 'label' in data:
for j in data['label']:
print(i, j, file=f)
else:
G_train = nx.Graph()
G_test = nx.Graph()
edges = np.random.permutation(list(Gs.edges()))
nodes = set()
for a, b in edges:
if a not in nodes or b not in nodes:
G_train.add_edge(a, b)
nodes.add(a)
nodes.add(b)
else:
G_test.add_edge(a, b)
assert len(nodes) == Gs.number_of_nodes()
assert len(nodes) == G_train.number_of_nodes()
num_test_edges = int((1 - radio) * Gs.number_of_edges())
now_number = G_test.number_of_edges()
if num_test_edges < now_number:
test_edges = list(G_test.edges())
G_train.add_edges_from(test_edges[:now_number -
num_test_edges])
G_test.remove_edges_from(test_edges[:now_number -
num_test_edges])
print(
"sample graph,origin: {} {}, train: {} {}, test: {} {}".format(
Gs.number_of_nodes(), Gs.number_of_edges(),
G_train.number_of_nodes(), G_train.number_of_edges(),
G_test.number_of_nodes(), G_test.number_of_edges()))
with utils.write_with_create(file_path) as f:
for i, j in G_train.edges():
print(i, j, file=f)
with utils.write_with_create(file_test_path) as f:
for i, j in G_test.edges():
print(i, j, file=f)