def __init__(self,
num_nodes,
edges,
node_types=None,
node_feat=None,
edge_feat=None):
self._g = hpg.HeterGraph(num_nodes, edges, node_types, node_feat,
edge_feat)
self._multi_graph = self._g._multi_graph
def build_graph(self):
"""Build pgl heterogeneous graph.
"""
self.conf_id2index, self.conf_name2index, conf_node_type = self.remapping_id(
self.config['data_path'] + 'id_conf.txt',
start_index=0,
node_type='conf')
logging.info('%d venues have been loaded.' % (len(self.conf_id2index)))
self.author_id2index, self.author_name2index, author_node_type = self.remapping_id(
self.config['data_path'] + 'id_author.txt',
start_index=len(self.conf_id2index),
node_type='author')
logging.info('%d authors have been loaded.' %
(len(self.author_id2index)))
self.paper_id2index, self.paper_name2index, paper_node_type = self.remapping_id(
self.config['data_path'] + 'paper.txt',
start_index=(len(self.conf_id2index) + len(self.author_id2index)),
node_type='paper',
separator='\t')
logging.info('%d papers have been loaded.' %
(len(self.paper_id2index)))
node_types = conf_node_type + author_node_type + paper_node_type
num_nodes = len(node_types)
edges_by_types = {}
paper_author_edges = self.load_edges(
self.config['data_path'] + 'paper_author.txt', self.paper_id2index,
self.author_id2index)
paper_conf_edges = self.load_edges(
self.config['data_path'] + 'paper_conf.txt', self.paper_id2index,
self.conf_id2index)
# edges_by_types['edge'] = paper_author_edges + paper_conf_edges
edges_by_types['p2c'] = paper_conf_edges
edges_by_types['c2p'] = [(dst, src) for src, dst in paper_conf_edges]
edges_by_types['p2a'] = paper_author_edges
edges_by_types['a2p'] = [(dst, src) for src, dst in paper_author_edges]
# logging.info('%d edges have been loaded.' %
# (len(edges_by_types['edge'])))
node_features = {
'index':
np.array([i for i in range(num_nodes)]).reshape(-1,
1).astype(np.int64)
}
self.graph = heter_graph.HeterGraph(num_nodes=num_nodes,
edges=edges_by_types,
node_types=node_types,
node_feat=node_features)
def build_graph(self):
"""Build pgl heterogeneous graph.
"""
edges_by_types = {}
npy = self.edge_file_list[0][1] + ".npy"
if os.path.exists(npy):
log.info("load data from numpy file")
for pair in self.edge_file_list:
edges_by_types[pair[0]] = np.load(pair[1] + ".npy")
else:
log.info("load data from txt file")
for pair in self.edge_file_list:
edges_by_types[pair[0]] = self.load_edges(pair[1])
# np.save(pair[1] + ".npy", edges_by_types[pair[0]])
for e_type, edges in edges_by_types.items():
log.info(["number of %s edges: " % e_type, len(edges)])
if self.symmetry:
tmp = {}
for key, edges in edges_by_types.items():
n_list = key.split('2')
re_key = n_list[1] + '2' + n_list[0]
tmp[re_key] = edges_by_types[key][:, [1, 0]]
edges_by_types.update(tmp)
log.info(["finished loadding symmetry edges."])
node_types = self.load_node_types(self.node_types_file)
assert len(node_types) == self.num_nodes, \
"num_nodes should be equal to the length of node_types"
log.info(["number of nodes: ", len(node_types)])
node_features = {
'index':
np.array([i for i in range(self.num_nodes)
]).reshape(-1, 1).astype(np.int64)
}
self.graph = heter_graph.HeterGraph(num_nodes=self.num_nodes,
edges=edges_by_types,
node_types=node_types,
node_feat=node_features)
def test_dump():
np.random.seed(1)
edges = {}
# for test no successor
edges['c2p'] = [(1, 4), (0, 5), (1, 9), (1, 8), (2, 8), (2, 5), (3, 6),
(3, 7), (3, 4), (3, 8)]
edges['p2c'] = [(v, u) for u, v in edges['c2p']]
edges['p2a'] = [(4, 10), (4, 11), (4, 12), (4, 14), (4, 13), (6, 12),
(6, 11), (6, 14), (7, 12), (7, 11), (8, 14), (9, 10)]
edges['a2p'] = [(v, u) for u, v in edges['p2a']]
node_types = ['c' for _ in range(4)] + ['p' for _ in range(6)
] + ['a' for _ in range(5)]
node_types = [(i, t) for i, t in enumerate(node_types)]
graph = heter_graph.HeterGraph(num_nodes=len(node_types),
edges=edges,
node_types=node_types)
graph.dump("./hetergraph_mmap", outdegree=True)
def build_graph(self):
"""Build pgl heterogeneous graph.
"""
edge_data_by_type, all_edges, all_nodes = self.load_training_data(
self.train_edges_file,
slf_loop=self.config['slf_loop'],
symmetry_edge=self.config['symmetry_edge'])
num_nodes = len(all_nodes)
node_features = {
'index':
np.array([i for i in range(num_nodes)],
dtype=np.int64).reshape(-1, 1)
}
self.graph = heter_graph.HeterGraph(num_nodes=num_nodes,
edges=edge_data_by_type,
node_types=None,
node_feat=node_features)
self.edge_types = sorted(self.graph.edge_types_info())
logging.info('total %d nodes are loaded' % (self.graph.num_nodes))
def setUpClass(cls):
np.random.seed(1)
edges = {}
# for test no successor
edges['c2p'] = [(1, 4), (0, 5), (1, 9), (1, 8), (2, 8), (2, 5), (3, 6),
(3, 7), (3, 4), (3, 8)]
edges['p2c'] = [(v, u) for u, v in edges['c2p']]
edges['p2a'] = [(4, 10), (4, 11), (4, 12), (4, 14), (4, 13), (6, 12),
(6, 11), (6, 14), (7, 12), (7, 11), (8, 14), (9, 10)]
edges['a2p'] = [(v, u) for u, v in edges['p2a']]
# for test speed
# edges['c2p'] = [(0, 4), (0, 5), (1, 9), (1,8), (2,8), (2,5), (3,6), (3,7), (3,4), (3,8)]
# edges['p2c'] = [(v,u) for u, v in edges['c2p']]
# edges['p2a'] = [(4,10), (4,11), (4,12), (4,14), (5,13), (6,13), (6,11), (6,14), (7,12), (7,11), (8,14), (9,13)]
# edges['a2p'] = [(v,u) for u, v in edges['p2a']]
node_types = ['c' for _ in range(4)] + ['p' for _ in range(6)
] + ['a' for _ in range(5)]
node_types = [(i, t) for i, t in enumerate(node_types)]
cls.graph = heter_graph.HeterGraph(
num_nodes=len(node_types), edges=edges, node_types=node_types)