def node2vec_embedding(graph, name):
rw = BiasedRandomWalk(graph)
walks = rw.run(graph.nodes(), n=num_walks, length=walk_length, p=p, q=q)
print(f"Number of random walks for '{name}': {len(walks)}")
model = Word2Vec(
walks,
size=dimensions,
window=window_size,
min_count=0,
sg=1,
workers=workers,
iter=num_iter,
)
def get_embedding(u):
#print(len(model.wv[u]),u)
if u == '16039':
return np.ndarray(128)
elif u == '24601':
return np.ndarray(128)
elif u == '21450':
return np.ndarray(128)
elif u == '12492':
return np.ndarray(128)
elif u == '6506':
return np.ndarray(128)
elif u == '1545':
return np.ndarray(128)
return model.wv[u]
return get_embedding
def walks(self, walklen, n1):
G = nx.Graph()
G = nx.read_weighted_edgelist(self.dataset + "/krnmdata1/CQAG1.txt")
rw = BiasedRandomWalk(StellarGraph(G))
weighted_walks = rw.run(
nodes=G.nodes(), # root nodes
length=walklen, # maximum length of a random walk
n=n1, # number of random walks per root node
p=0.5, # Defines (unormalised) probability, 1/p, of returning to source node
q=2.0, # Defines (unormalised) probability, 1/q, for moving away from source node
weighted=True, #for weighted random walks
seed=42 # random seed fixed for reproducibility
)
print("Number of random walks: {}".format(len(weighted_walks)))
#print(weighted_walks[0:10])
#remove answer nodes
walks = []
for i in range(len(weighted_walks)):
walk = weighted_walks[i]
w = []
for node in walk:
if int(node) < self.qnum:
w.append(node)
elif int(node) > (self.qnum + self.anum):
n = int(node) - self.anum
w.append(str(n))
walks.append(w)
print(walks[0:10])
return walks
def learn_embeddings(self,
embedding_dim=100,
window_size=5,
max_rw_len=50,
walks_per_node=20,
p=0.5,
q=2.0):
print('Running node2vec...')
rw = BiasedRandomWalk(StellarGraph(self.graph))
walks = rw.run(nodes=list(self.graph),
length=max_rw_len,
n=walks_per_node,
p=p,
q=q)
print(f'Number of random walks: {len(walks)}')
print('Running word2vec...')
model = Word2Vec(walks,
size=embedding_dim,
window=window_size,
min_count=0,
sg=1,
workers=2,
iter=1)
model.init_sims(replace=True)
return model.wv
def node2vec_embedding(graph, name, weighted=False):
p = 1.0
q = 1.0
dimensions = 128
num_walks = 10
walk_length = 80
window_size = 10
num_iter = 1
workers = multiprocessing.cpu_count()
rw = BiasedRandomWalk(graph)
walks = rw.run(graph.nodes(),
n=num_walks,
length=walk_length,
p=p,
q=q,
weighted=weighted)
print(f"Number of random walks for '{name}': {len(walks)}")
model = Word2Vec(walks, size=dimensions, window=window_size, min_count=0,
sg=1, workers=workers, iter=num_iter)
def get_embedding(u):
return model.wv[u]
return get_embedding
def node2vec_embedding(graph):
p = 1.0
q = 1.0
dimensions = 128
num_walks = 10
walk_length = 80
window_size = 10
num_iter = 1
workers = multiprocessing.cpu_count()
graph = StellarGraph(graph)
rw = BiasedRandomWalk(graph)
walks = rw.run(graph.nodes(), n=num_walks, length=walk_length, p=p, q=q)
print(f"Number of random walks: {len(walks)}")
model = Word2Vec(walks,
size=dimensions,
window=window_size,
min_count=0,
sg=1,
workers=workers,
iter=num_iter)
features = pd.DataFrame(data=model.wv.vectors, index=list(graph.nodes()))
features.index = features.index.map(str)
return features
def graph_embed():
combine = get_combine()
li = ['bank', 'acquirer', 'coin', 'mcc', 'shop', 'nation', 'city']
d = {
'bank': 'b',
'mcc': 'm',
'acquirer': 'a',
'coin': 'c',
'shop': 's',
'nation': 'n',
'city': 'z'
}
have_df = False
df_all = None
for col_a in li:
combine[col_a] = combine[col_a].astype(str) + [d[col_a]]
for index, col_a in enumerate(li[1:]):
print(f'{col_a} started..')
walk_all = []
for day in np.linspace(1, 120, 120):
print(day, end=',', flush=True)
df = combine[combine['date'] == day]
G = construct_graph('bank', col_a, df)
rw = BiasedRandomWalk(StellarGraph(G))
walk = rw.run(
nodes=list(G.nodes()), # root nodes
length=80, # maximum length of a random walk
n=1, # number of random walks per root node
p=1, # Defines (unormalised) probability, 1/p, of returning to source node
q=1, # Defines (unormalised) probability, 1/q, for moving away from source node
)
walk_all.extend(walk)
del df, G, rw, walk
gc.collect()
model = Word2Vec(walk_all,
size=5,
window=3,
min_count=1,
sg=0,
workers=16,
iter=10)
temp_d = {}
for w in list(model.wv.vocab):
temp_d[w] = model[w]
temp_df = pd.DataFrame(
data=combine[col_a].map(temp_d).tolist(),
columns=['embed_bank_' + col_a + str(x + 1) for x in range(5)])
if (have_df):
df_all = pd.concat([df_all, temp_df], axis=1)
else:
df_all = temp_df
have_df = True
del temp_d, model
gc.collect()
return df_all
def _fit_node2vec(train_graph, params, edge_weight=None):
rw = BiasedRandomWalk(train_graph)
walks = rw.run(
nodes=list(train_graph.nodes()), # root nodes
length=params["length"],
n=params["number_of_walks"],
p=params["random_walk_p"],
q=params["random_walk_q"],
weighted=edge_weight is not None
)
model = Word2Vec(walks, size=params["embedding_dimension"])
return model.wv[train_graph.nodes()]
def walks(self,walklen):
G=nx.Graph();
G=nx.read_weighted_edgelist(self.dataset+"/krnmdata1/teamsG.txt")
rw = BiasedRandomWalk(StellarGraph(G))
weighted_walks = rw.run(
nodes=G.nodes(), # root nodes
length=walklen, # maximum length of a random walk
n=5, # number of random walks per root node
p=0.1, # Defines (unormalised) probability, 1/p, of returning to source node
q=2.0, # Defines (unormalised) probability, 1/q, for moving away from source node
weighted=True, #for weighted random walks
seed=42 # random seed fixed for reproducibility
)
print("Number of random walks: {}".format(len(weighted_walks)))
print(weighted_walks[0:10])
return weighted_walks
def node2vec_embedding(graph, name):
rw = BiasedRandomWalk(graph)
walks = rw.run(graph.nodes(), n=num_walks, length=walk_length, p=p, q=q)
print(f"Number of random walks for '{name}': {len(walks)}")
model = Word2Vec(
walks,
size=dimensions,
window=window_size,
min_count=0,
sg=1,
workers=workers,
iter=num_iter,
)
def get_embedding(u):
return model.wv[u]
return get_embedding
def node2vec_walk(G, params):
"""Performs biased random walks using StellarGraph to generate corpus used in node2vec and writes corpus to a txt file
:param G : StellarGraph graph
Nodes consist of apps, api calls, packages, and invoke methods
:param params : dict
dict["key"] where dict is global parameter dictionary and key returns node2vec parameter sub-dictionary
"""
start_walks = time.time()
print("Starting Random Walks")
rw = BiasedRandomWalk(G)
fp = os.path.join(params["save_dir"], params["filename"])
os.makedirs(params["save_dir"], exist_ok=True)
walks = rw.run(
nodes=list(G.nodes(node_type="app_nodes")), # root nodes
length=params["length"], # maximum length of a random walk
n=params["n"], # number of random walks per root node
p=params["p"], # Defines prob, 1/p, of returning to source node
q=params["q"], # Defines prob, 1/q, for moving away from source node
)
print("--- Done Walking in " + str(int(time.time() - start_walks)) +
" Seconds ---")
print()
print("Number of random walks: {}".format(len(walks)))
# save walks to file
with open(fp, 'w') as f:
for walk in walks:
for node in walk:
f.write(str(node) + ' ')
f.write('\n')
f.close()
if params["verbose"]:
print("Saved %s to %s" % (params["filename"], params["save_dir"]))
return
def get_node_feats(adj): # input is cur_adj
edgelist = adj['idx'].cpu().data.numpy()
source = edgelist[:, 0]
target = edgelist[:, 1]
weight = np.ones(len(source))
G = pd.DataFrame({
'source': source,
'target': target,
'weight': weight
})
G = StellarGraph(edges=G)
rw = BiasedRandomWalk(G)
weighted_walks = rw.run(
nodes=list(G.nodes()), # root nodes
length=2, # maximum length of a random walk
n=5, # number of random walks per root node
p=1, # Defines (unormalised) probability, 1/p, of returning to source node
q=0.5, # Defines (unormalised) probability, 1/q, for moving away from source node
weighted=True, # for weighted random walks
seed=42, # random seed fixed for reproducibility
)
str_walks = [[str(n) for n in walk] for walk in weighted_walks]
weighted_model = Word2Vec(str_walks,
size=self.feats_per_node,
window=5,
min_count=0,
sg=1,
workers=1,
iter=1)
# Retrieve node embeddings and corresponding subjects
node_ids = weighted_model.wv.index2word # list of node IDs
# change to integer
for i in range(0, len(node_ids)):
node_ids[i] = int(node_ids[i])
weighted_node_embeddings = (
weighted_model.wv.vectors
) # numpy.ndarray of size number of nodes times embeddings dimensionality
# create dic
dic = dict(zip(node_ids, weighted_node_embeddings.tolist()))
# ascending order
dic = dict(sorted(dic.items()))
# create matrix
adj_mat = sp.lil_matrix((self.data.num_nodes, self.feats_per_node))
for row_idx in node_ids:
adj_mat[row_idx, :] = dic[row_idx]
adj_mat = adj_mat.tocsr()
adj_mat = adj_mat.tocoo()
coords = np.vstack((adj_mat.row, adj_mat.col)).transpose()
values = adj_mat.data
row = list(coords[:, 0])
col = list(coords[:, 1])
indexx = torch.LongTensor([row, col])
tensor_size = torch.Size(
[self.data.num_nodes, self.feats_per_node])
degs_out = torch.sparse.FloatTensor(indexx,
torch.FloatTensor(values),
tensor_size)
hot_1 = {
'idx': degs_out._indices().t(),
'vals': degs_out._values()
}
return hot_1
def node2vec():
print('Training Node2Vec mode!')
# initialize results arrays
total_mse = np.zeros(args.exp_number)
total_pcc = np.zeros(args.exp_number)
total_mae = np.zeros(args.exp_number)
mse_datasets = {}
std_datasets = {}
pcc_datasets = {}
pcc_std_datasets = {}
mae_datasets = {}
mae_std_datasets = {}
t_total = time.time()
if args.dataset == 'all':
datasets = [
'airport', 'collaboration', 'congress', 'forum', 'geom', 'astro'
]
else:
datasets = [args.dataset]
for dataset in datasets:
for exp_number in range(args.exp_number):
print("%s: experiment number %d" % (dataset, exp_number + 1))
data = preprocess_dataset.clean_data(dataset)
if dataset != 'usair':
data['weights'] = preprocessing.normalize([data['weights']])[0]
# random split of data
data_train, data_test = train_test_split(data, test_size=0.2)
data_train, data_val = train_test_split(data_train, test_size=0.08)
data = data.reset_index()
data_train = data_train.reset_index()
data_val = data_val.reset_index()
data_test = data_test.reset_index()
G = preprocess_dataset.create_graph_gcn(dataset, data, data_train)
val_G = preprocess_dataset.create_graph_gcn(
dataset, data, data_val)
test_G = preprocess_dataset.create_graph_gcn(
dataset, data, data_test)
nodes_len = len(G.nodes)
node_ids_to_index = {}
for i, node_id in enumerate(G.nodes):
node_ids_to_index[node_id] = i
train_A = nx.adjacency_matrix(G)
val_A = nx.adjacency_matrix(val_G)
test_A = nx.adjacency_matrix(test_G)
train_labels = torch.FloatTensor(
data_train['weights'].values).cuda()
val_labels = torch.FloatTensor(data_val['weights'].values).cuda()
test_labels = torch.FloatTensor(data_test['weights'].values).cuda()
train_A = sparse_mx_to_torch_sparse_tensor(train_A).cuda()
val_A = sparse_mx_to_torch_sparse_tensor(val_A).cuda()
test_A = sparse_mx_to_torch_sparse_tensor(test_A).cuda()
G = sg.from_networkx(G)
rw = BiasedRandomWalk(G)
weighted_walks = rw.run(
nodes=G.nodes(), # root nodes
length=args.length, # maximum length of a random walk
n=args.n_size, # number of random walks per root node
p=args.
p, # Defines (unormalised) probability, 1/p, of returning to source node
q=args.
q, # Defines (unormalised) probability, 1/q, for moving away from source node
weighted=True, # for weighted random walks
seed=42, # random seed fixed for reproducibility
)
print("Number of random walks: {}".format(len(weighted_walks)))
weighted_model = Word2Vec(weighted_walks,
vector_size=args.vector_size,
window=5,
min_count=0,
sg=1,
workers=4)
weights = torch.FloatTensor(weighted_model.wv.vectors).cuda()
########################################
train_n1 = torch.tensor(data_train['A'].values).cuda()
train_n2 = torch.tensor(data_train['B'].values).cuda()
train_n1_indices = torch.ones(train_n1.shape[0])
for i, value in enumerate(train_n1):
train_n1_indices[i] = node_ids_to_index[value.item()]
train_n1_indices = train_n1_indices.cuda().long()
train_n2_indices = torch.ones(train_n1.shape[0])
for i, value in enumerate(train_n2):
train_n2_indices[i] = node_ids_to_index[value.item()]
train_n2_indices = train_n2_indices.cuda().long()
########################################
val_n1 = torch.tensor(data_val['A'].values).cuda()
val_n2 = torch.tensor(data_val['B'].values).cuda()
val_n1_indices = torch.ones(val_n1.shape[0])
for i, value in enumerate(val_n1):
val_n1_indices[i] = node_ids_to_index[value.item()]
val_n1_indices = val_n1_indices.cuda().long()
val_n2_indices = torch.ones(val_n1.shape[0])
for i, value in enumerate(val_n2):
val_n2_indices[i] = node_ids_to_index[value.item()]
val_n2_indices = val_n2_indices.cuda().long()
########################################
test_n1 = torch.tensor(data_test['A'].values).cuda()
test_n2 = torch.tensor(data_test['B'].values).cuda()
test_n1_indices = torch.ones(test_n1.shape[0])
for i, value in enumerate(test_n1):
test_n1_indices[i] = node_ids_to_index[value.item()]
test_n1_indices = test_n1_indices.cuda().long()
test_n2_indices = torch.ones(test_n1.shape[0])
for i, value in enumerate(test_n2):
test_n2_indices[i] = node_ids_to_index[value.item()]
test_n2_indices = test_n2_indices.cuda().long()
########################################
model = Node2Vec(weights, 0.5)
optimizer = optim.Adam(model.parameters(), lr=args.lr)
model.train()
model = model.to(args.device)
# train
for epoch in range(args.epochs):
t = time.time()
model.train()
optimizer.zero_grad()
output = model(train_n1_indices, train_n2_indices)
loss_train = F.mse_loss(output, train_labels)
loss_train.backward()
optimizer.step()
# validation
model.eval()
output = model(val_n1_indices, val_n2_indices)
loss_val = F.mse_loss(output, val_labels)
if args.verbose:
print('Epoch: {:04d}'.format(epoch + 1),
'loss_train: {:.4f}'.format(loss_train.item()),
'loss_val: {:.4f}'.format(loss_val.item()),
'time: {:.4f}s'.format(time.time() - t))
# test
model.eval()
with torch.no_grad():
output = model(test_n1_indices, test_n2_indices)
loss_test = F.mse_loss(torch.flatten(output), test_labels)
pcc_test = pearson_correlation(test_labels, output)
mae_test = F.l1_loss(output, test_labels)
print("Test set results:",
"loss= {:.10f}".format(loss_test.item()),
"pcc= {:.10f}".format(pcc_test),
"mae= {:.10f}".format(mae_test.item()))
total_mse[exp_number] = loss_test
total_pcc[exp_number] = pcc_test
total_mae[exp_number] = mae_test
# results
mse_datasets[dataset] = np.mean(total_mse)
std_datasets[dataset] = np.std(total_mse)
total_mse = np.zeros(args.exp_number)
pcc_datasets[dataset] = np.mean(total_pcc[~np.isnan(total_pcc)])
pcc_std_datasets[dataset] = np.std(total_pcc[~np.isnan(total_pcc)])
total_pcc = np.zeros(args.exp_number)
mae_datasets[dataset] = np.mean(total_mae)
mae_std_datasets[dataset] = np.std(total_mae)
total_mae = np.zeros(args.exp_number)
for dataset in datasets:
print("MSE %s: {:,f}".format(mse_datasets[dataset]) % dataset)
print("MSE_STD %s: {:,f}".format(std_datasets[dataset]) % dataset)
print("PCC %s: {:,f}".format(pcc_datasets[dataset]) % dataset)
print("PCC_STD %s: {:,f}".format(pcc_std_datasets[dataset]) % dataset)
print("MAE %s: {:,f}".format(mae_datasets[dataset]) % dataset)
print("MAE_STD %s: {:,f}".format(mae_std_datasets[dataset]) % dataset)
print("Total time elapsed: {:.4f}s".format(time.time() - t_total))
exit()
save_path = project_folder + "/" + use_model_type + "_" + uni_name + ".csv"
G_graphml = nx.read_graphml(load_path)
# get the node features as a dataframe, these will then be added to the stellar graph.
# This seems to work better than trying to put them in directly
# nodefeatures = pd.DataFrame.from_dict(dict(G_graphml.nodes(data=True)), orient='index')
# print(nodefeatures)
# Convert the networkx graph to a Stellargraph
G = StellarGraph.from_networkx(G_graphml)
rw = BiasedRandomWalk(G)
walks = rw.run(
nodes=list(G.nodes()), # root nodes
length=30, # maximum length of a random walk
n=100, # number of random walks per root node
p=0.5, # Defines (unormalised) probability, 1/p, of returning to source node
q=2.0, # Defines (unormalised) probability, 1/q, for moving away from source node
)
print("Number of random walks: {}".format(len(walks)))
str_walks = [[str(n) for n in walk] for walk in walks]
model = Word2Vec(str_walks,
size=dims,
window=10,
min_count=0,
sg=1,
workers=1,
iter=1)
model.wv.save_word2vec_format(save_path)