def get_robustness_zero(dataset):
for crawler_type in crawler_list:
robustness_zero[crawler_type] = dict()
for i in range(0, MAX_GEN):
robustness_zero[crawler_type][i] = dict()
ad_p_remove = "0"
s1_fname = directory + str(ad_p_remove) + '/' + \
dataset + '/' + crawler_type + '/' + \
dataset + '_' + str(i + 1) + '_1'
sample_graph_1 = _mylib.read_file(s1_fname)
node_sim_list = []
node_cov_sim_list = []
partition_sim_list = []
degree_sim_list = []
for j in range(0, MAX_GEN):
if i == j:
continue
s2_fname = directory + str(ad_p_remove) + '/' + \
dataset + '/' + crawler_type + '/' + \
dataset + '_' + str(j + 1) + '_1'
sample_graph_2 = _mylib.read_file(s2_fname)
sample_robustness = calculateRobustness(
sample_graph_1, sample_graph_2, s1_fname, s2_fname)
node_sim_list.append(sample_robustness['node_sim'])
node_cov_sim_list.append(
sample_robustness['node_coverage_sim'])
partition_sim_list.append(sample_robustness['partition_sim'])
degree_sim_list.append(sample_robustness['degree_sim'])
robustness_zero[crawler_type][i]['node_sim'] = 1. * sum(
node_sim_list) / len(node_sim_list)
robustness_zero[crawler_type][i]['node_cov_sim'] = 1. * sum(
node_cov_sim_list) / len(node_cov_sim_list)
robustness_zero[crawler_type][i]['partition_sim'] = 1. * sum(
partition_sim_list) / len(partition_sim_list)
robustness_zero[crawler_type][i]['degree_sim'] = 1. * sum(
degree_sim_list) / len(degree_sim_list)
return robustness_zero
dataset = args.dataset
directory = './data-adv/rand_edge/'
# dataset = 'socfb-Amherst41'
log_file = './log/' + dataset
MAX_GEN = 10
ad_p_remove = "0"
crawler_list = ['med', 'mod', 'rw', 'bfs', 'rand']
# crawler_list = ['mod']
for crawler_type in crawler_list:
for i in range(0, MAX_GEN):
fname = directory + str(ad_p_remove) + '/' + \
dataset + '/' + crawler_type + '/' + \
dataset + '_' + str(i+1) + '_1'
G = _mylib.read_file(fname)
node_count = G.number_of_nodes()
com_G = get_community(G, fname)
node_G = G.nodes()
deg_hist_G = nx.degree_histogram(G)
for j in range(0, MAX_GEN):
if i == j:
continue
fname = directory + str(ad_p_remove) + '/' + \
dataset + '/' + crawler_type + '/' + \
dataset + '_' + str(j + 1) + '_1'
G = _mylib.read_file(fname)
elif remove_type == 'edge_incomp':
directory = './data-adv/edge_incomp/'
nodes_original = set()
log_file = './log/' + dataset + '_' + remove_type
MAX_GEN = 10
# crawler_list = ['med','mod','rw', 'bfs', 'rand']
crawler_list = ['mod', 'rw', 'bfs', 'rand']
for crawler_type in crawler_list:
for i in range(0, MAX_GEN):
s1_fname = directory + '0' + '/' + \
dataset + '/' + crawler_type + '/' + \
dataset + '_' + str(i + 1) + '_1'
sample_graph_1 = _mylib.read_file(s1_fname)
# for ad_p_remove in ["0.01", "0.05", "0.1", "0.2", "0.3"]:
for ad_p_remove in ["0.1", "0.2", "0.3", "0.4", "0.5"]:
s2_fname = directory + ad_p_remove + '/' + \
dataset + '/' + crawler_type + '/' + \
dataset + '_' + str(i + 1) + '_1'
sample_graph_2 = _mylib.read_file(s2_fname)
deleted_nodes_count = int(
float(ad_p_remove) * len(nodes_original))
# print(' Delete nodes count: {}/{} \t {}'.format(deleted_nodes_count, len(nodes_original), ad_p_remove))
partition_sim = calculate_community_sim(
sample_graph_1, sample_graph_2, s1_fname, s2_fname,
deleted_nodes_count)
parser.add_argument('fname', help='Edgelist file', type=str)
parser.add_argument('-remove',
help='remove node or edge',
type=str,
default='edge')
args = parser.parse_args()
fname = args.fname
remove_type = args.remove
fname = fname.replace('\\', '/')
network_name = fname.split('.')[1].split('/')[-1]
log_file = './log/eigen_' + remove_type + '.txt'
G = _mylib.read_file(fname, isDirected=False)
e = compute_eigenvals(G, network_name)
max_e = e[-1]
print(max_e)
# for ad_p_remove in [0.01, 0.05, 0.1, 0.2, 0.3]:
# e_list = list()
# for i in range(0, 10):
# if remove_type == 'edge':
# directory = './data-adv/rand_edge/' + str(ad_p_remove) + '/'
# elif remove_type == 'node':
# directory = './data-adv/rand_node/' + str(ad_p_remove) + '/'
# elif remove_type == 'target':
# directory = './data-adv/target/' + str(ad_p_remove) + '/'
#
directory = './data-adv/rand_node/'
nodes_original = pickle.load(open(directory + dataset + '_nodes.pickle'))
elif remove_type == 'target':
directory = './data-adv/target/'
nodes_original = pickle.load(open(directory + dataset + '_nodes.pickle'))
elif remove_type == 'edge_incomp':
directory = './data-adv/edge_incomp/'
nodes_original = set()
log_file = './log/' + dataset + '_' + remove_type
MAX_GEN = 10
# crawler_list = ['med','mod','rw', 'bfs', 'rand']
crawler_list = ['mod', 'rw', 'bfs', 'rand']
G_original = _mylib.read_file(fname)
deg = dict(G_original.degree())
avg_deg = np.average(deg.values())
e = adv_norm.compute_eigenvals(G_original, dataset)
max_e = e[-1]
cc = nx.average_clustering(G_original)
for crawler_type in crawler_list:
for i in range(0, MAX_GEN):
s1_fname = directory + '0' + '/' + \
dataset + '/' + crawler_type + '/' + \
dataset + '_' + str(i + 1) + '_1'
sample_graph_1 = _mylib.read_file(s1_fname)
deg = dict(sample_graph_1.degree())
avg_deg_0 = np.average(deg.values())