def random_walk_multi_layers(multi_layers_net,
walk_iters=100,
walk_legth=160,
save_random_walk=False):
'''
:param multi_layers_net: list，表示一个多层的大网络，每一层都是一个小网络，并且每一层的节点都相同
:param output_file: str，表示输出文件的路径，文件内容为节点及其对应的向量
:param walk_iters: int，表示单个节点游走的次数
:param walk_legth: int，表示一个节点在网络中游走的步数
:param save_random_walk: boolean，表示是否保存随机游走的路径
:return:
'''
multi_layers = defaultdict()
for i in range(len(multi_layers_net)):
G = read_graph(multi_layers_net[i], weighted=True)
print("{} layer -> {} nodes and {} edges.".format(
i, len(nx.nodes(G)), len(nx.edges(G))))
multi_layers[i] = G
nodes = multi_layers[0].nodes()
walks = []
for index, node in enumerate(nodes):
time1 = time.clock()
max_weights = list()
for key, G in multi_layers.items():
nei_weight = set()
for nei_node in G.neighbors(node):
nei_weight.add(float(G[node][nei_node]['weight']))
max_weights.append(max(nei_weight))
select_layer = max_weights.index(max(max_weights))
for walk_iter in range(walk_iters):
walk = random_walk(multi_layers[select_layer], node, walk_legth)
walks.append(walk)
time2 = time.clock()
print("{} * {} -> {} layer: cost {}s".format(index, node,
select_layer + 1,
time2 - time1))
if save_random_walk:
FileUtil.write2DemList2File(walks, "./random_walk.txt")
return walks
def random_walk(edges, num_walks=100, walk_length=160):
time1 = time.clock()
nx_G = read_graph(edges, weighted=True, directed=False)
print("{} nodes, {} edges.".format(len(nx.nodes(nx_G)),
len(nx.edges(nx_G))))
time2 = time.clock()
print("It cost {}s to read edges.".format(time2 - time1))
G = Graph(nx_G, p=1, q=1)
print("generate transition matrix......")
G.preprocess_transition_probs()
time3 = time.clock()
print("It cost {}s to generate transition matrix.".format(time3 - time2))
print("begin to random walk......")
walks = G.simulate_walks(num_walks=num_walks, walk_length=walk_length)
time4 = time.clock()
print("It cost {}s to random walk.".format(time4 - time3))
FileUtil.write2DemList2File(walks, "./node2vec_walks.txt")
return walks
