def load(dataset):
datadir = os.path.join('data', dataset)
if not os.path.exists(datadir):
os.makedirs(datadir)
ds = download(dataset)
adj = nx.to_numpy_array(ds.graph)
diff = compute_ppr(ds.graph, 0.2)
feat = ds.features[:]
labels = ds.labels[:]
idx_train = np.argwhere(ds.train_mask == 1).reshape(-1)
idx_val = np.argwhere(ds.val_mask == 1).reshape(-1)
idx_test = np.argwhere(ds.test_mask == 1).reshape(-1)
np.save(f'{datadir}/adj.npy', adj)
np.save(f'{datadir}/diff.npy', diff)
np.save(f'{datadir}/feat.npy', feat)
np.save(f'{datadir}/labels.npy', labels)
np.save(f'{datadir}/idx_train.npy', idx_train)
np.save(f'{datadir}/idx_val.npy', idx_val)
np.save(f'{datadir}/idx_test.npy', idx_test)
else:
adj = np.load(f'{datadir}/adj.npy')
diff = np.load(f'{datadir}/diff.npy')
feat = np.load(f'{datadir}/feat.npy')
labels = np.load(f'{datadir}/labels.npy')
idx_train = np.load(f'{datadir}/idx_train.npy')
idx_val = np.load(f'{datadir}/idx_val.npy')
idx_test = np.load(f'{datadir}/idx_test.npy')
if dataset == 'citeseer':
feat = preprocess_features(feat)
epsilons = [1e-5, 1e-4, 1e-3, 1e-2]
avg_degree = np.sum(adj) / adj.shape[0]
epsilon = epsilons[np.argmin([
abs(avg_degree - np.argwhere(diff >= e).shape[0] / diff.shape[0])
for e in epsilons
])]
diff[diff < epsilon] = 0.0
scaler = MinMaxScaler()
scaler.fit(diff)
diff = scaler.transform(diff)
ori_adj = copy.deepcopy(adj)
# print(ori_adj)
adj = normalize_adj(adj + sp.eye(adj.shape[0])).todense()
return ori_adj, adj, diff, feat, labels, idx_train, idx_val, idx_test
def process_raw_data(self):
root = os.path.abspath(
os.path.join(os.path.dirname(__file__), os.path.pardir))
prefix = os.path.join(root, 'dataset', self.__graph_type, self.__name,
self.__name)
print(prefix)
graph_node_dict = {}
with open('{0}_graph_indicator.txt'.format(prefix), 'r') as f:
for idx, line in enumerate(f):
graph_node_dict[idx + 1] = int(line.strip('\n'))
max_nodes = Counter(graph_node_dict.values()).most_common(1)[0][1]
node_labels = []
if os.path.exists('{0}_node_labels.txt'.format(prefix)):
with open('{0}_node_labels.txt'.format(prefix), 'r') as f:
for line in f:
node_labels += [int(line.strip('\n'))] # -1 ???????????
num_unique_node_labels = max(node_labels) + 1 # ??????????????
else:
print('No node labels')
node_attrs = []
if os.path.exists('{0}_node_attributes.txt'.format(prefix)):
with open('{0}_node_attributes.txt'.format(prefix), 'r') as f:
for line in f:
node_attrs.append(
np.array([
float(attr)
for attr in re.split("[,\s]+", line.strip('\s\n'))
if attr
],
dtype=np.float))
else:
print('No node attributes')
graph_labels = []
unique_labels = set()
with open('{0}_graph_labels.txt'.format(prefix), 'r') as f:
for line in f:
val = int(line.strip('\n'))
if val not in unique_labels:
unique_labels.add(val)
graph_labels.append(val)
label_idx_dict = {val: idx for idx, val in enumerate(unique_labels)}
graph_labels = np.array(
[label_idx_dict[element] for element in graph_labels])
adj_list = {idx: []
for idx in range(1,
len(graph_labels) + 1)} # 每个图包含的边
index_graph = {idx: []
for idx in range(1,
len(graph_labels) + 1)} # 每个图包含的节点
with open('{0}_A.txt'.format(prefix), 'r') as f:
for line in f:
u, v = tuple(map(int, line.strip('\n').split(',')))
adj_list[graph_node_dict[u]].append((u, v))
index_graph[graph_node_dict[u]] += [u, v]
for k in index_graph.keys():
index_graph[k] = [u - 1 for u in set(index_graph[k])]
graphs, aug_view = [], []
for idx in range(1, 1 + len(adj_list)):
graph = nx.from_edgelist(adj_list[idx])
if max_nodes is not None and graph.number_of_nodes() > max_nodes:
continue
graph.graph['label'] = graph_labels[idx - 1]
for u in graph.nodes():
if len(node_labels) > 0:
node_label_one_hot = [0] * num_unique_node_labels
node_label = node_labels[u - 1]
node_label_one_hot[node_label] = 1
graph.nodes[u]['label'] = node_label_one_hot
if len(node_attrs) > 0:
graph.nodes[u]['feat'] = node_attrs[u - 1]
if len(node_attrs) > 0:
graph.graph['feat_dim'] = node_attrs[0].shape[0]
# relabeling
mapping = {}
for node_idx, node in enumerate(graph.nodes()):
mapping[node] = node_idx
graphs.append(nx.relabel_nodes(graph, mapping))
aug_view.append(compute_ppr(graph, alpha=0.2))
if 'feat_dim' in graphs[0].graph:
pass
else:
max_deg = max(
[max(dict(graph.degree).values()) for graph in graphs])
for graph in graphs:
for u in graph.nodes(data=True):
f = np.zeros(max_deg + 1)
f[graph.degree[u[0]]] = 1.0
if 'label' in u[1]:
f = np.concatenate((np.array(u[1]['label'],
dtype=np.float), f))
graph.nodes[u[0]]['feat'] = f
return graphs, aug_view
return 0
if __name__ == '__main__':
import warnings
warnings.filterwarnings("ignore")
a = 1
# print(os.path.abspath(os.path.join(os.path.dirname(__file__), os.path.pardir)))
ds = CitationGraphDataset(name='pubmed')
# nx.draw(ds.graph, node_size=10)
# plt.show()
# print(nx.to_numpy_array(ds.graph).shape)
print('start')
start = time.time()
# spectral_perturbation(ds.graph)
compute_ppr(ds.graph, 0.2)
finish = time.time()
print('finish ', finish - start)
# print(ds)
# print(nx.to_numpy_array(ds.graph).shape)
# print(n)
# print(ds.features)
# path = 'D:/VSrepos/spectral-graph-augmentation/dataset/node'
# if not os.path.exists(path):
# print('create')
# os.makedirs(path)
# else:
# print('exist')
# GraphData('PTC_MR').process_raw_data()