def __init__(self, model: Model, training_set_path=None, test_set_path=None, nodes_path=None, mappings_avail=False):
self.model = model
if training_set_path:
self.training_examples = pandas.read_csv(training_set_path, sep="\t", names=globConst.COL_NAMES_SAMPLES)
else:
self.training_examples = pandas.DataFrame(columns=globConst.COL_NAMES_SAMPLES)
if test_set_path:
self.test_examples = pandas.read_csv(test_set_path, sep="\t", names=globConst.COL_NAMES_SAMPLES)
else:
self.test_examples = pandas.DataFrame(columns=globConst.COL_NAMES_SAMPLES)
if nodes_path is not None:
self.nodes = pandas.read_csv(nodes_path, sep="\t", names=globConst.COL_NAMES_NODES)
else:
self.nodes = None
if not mappings_avail:
relation_labels = set()
relation_labels.update(set(self.test_examples[globConst.EDGE_TYPE_COL_NAME]))
relation_labels.update(set(self.training_examples[globConst.EDGE_TYPE_COL_NAME]))
self.node_label_to_id = None
self.node_types_to_id = None
self.relation_label_to_id = utils.create_mappings(relation_labels)
if self.nodes is not None:
self.node_label_to_id = utils.create_mappings(np.unique(self.nodes.values[:, 0]))
self.node_types_to_id = utils.create_mappings(np.unique(self.nodes.values[:, 1]))
else:
node_labels = set()
node_labels.update(set(self.test_examples[globConst.NODE1_ID_COL_NAME]))
node_labels.update(set(self.test_examples[globConst.NODE2_ID_COL_NAME]))
node_labels.update(set(self.training_examples[globConst.NODE1_ID_COL_NAME]))
node_labels.update(set(self.training_examples[globConst.NODE2_ID_COL_NAME]))
self.node_label_to_id = utils.create_mappings(node_labels)
# output mappings
io.write_mappings(
node_label_to_id=self.node_label_to_id,
node_types_to_id=self.node_types_to_id,
relation_label_to_id=self.relation_label_to_id,
)
else:
# testme
output_directory = os.path.join(
os.path.join(globConst.WORKING_DIR, evalConst.EVAL_OUTPUT_FOLDER_NAME), evalConst.MODEL_DIR
)
self.node_label_to_id = io.read_mapping(
os.path.join(output_directory, evalConst.MODEL_ENTITY_NAME_MAPPING_NAME)
)
node_type_path = os.path.join(output_directory, evalConst.MODEL_ENTITY_TYPE_MAPPING_NAME)
if os.path.exists(node_type_path):
self.node_types_to_id = io.read_mapping(node_type_path)
elif self.nodes is not None:
self.node_types_to_id = utils.create_mappings(np.unique(self.nodes.values[:, 1]))
io.write_mappings(node_types_to_id=self.node_types_to_id)
else:
pass # fixme error nodes must be provided when eval
self.relation_label_to_id = io.read_mapping(
os.path.join(output_directory, evalConst.MODEL_RELATION_TYPE_MAPPING_NAME)
)
def create_mappings(self, relations, node_types, node_labels=None):
self.node_label_to_id = utils.create_mappings(node_labels)
self.node_types_to_id = utils.create_mappings(node_types)
self.relation_label_to_id = utils.create_mappings(relations)