import os
import numpy as np
import scipy.sparse as sp
from kgcnn.data.base import MemoryGraphDataset
from kgcnn.data.download import DownloadDataset
from kgcnn.graph.methods import convert_scaled_adjacency_to_list
[docs]class CoraDataset(DownloadDataset, MemoryGraphDataset):
r"""Store and process (full) Cora dataset.
Data loaded from `<https://github.com/abojchevski/graph2gauss/raw/master/data>`_ .
From Paper: `"Deep Gaussian Embedding of Graphs: Unsupervised Inductive Learning via
Ranking" <https://arxiv.org/abs/1707.03815>`_ paper.
Nodes represent documents and edges represent citation links.
References:
(1) Bojchevski, Aleksandar and Günnemann, Stephan, Deep Gaussian Embedding of Graphs: Unsupervised Inductive
Learning via Ranking, arXiv, doi:10.48550/ARXIV.1707.03815, 2017
"""
download_info = {
"dataset_name": "Cora",
"data_directory_name": "cora",
"download_url": "https://github.com/abojchevski/graph2gauss/raw/master/data/cora.npz",
"download_file_name": 'cora.npz',
"unpack_tar": False,
"unpack_zip": False,
"unpack_directory_name": None
}
[docs] def __init__(self, reload=False, verbose: int = 10):
"""Initialize full Cora dataset.
Args:
reload (bool): Whether to reload the data and make new dataset. Default is False.
verbose (int): Print progress or info for processing where 60=silent. Default is 10.
"""
# Use default base class init()
self.data_keys = None
MemoryGraphDataset.__init__(self, dataset_name="Cora", verbose=verbose)
DownloadDataset.__init__(self, **self.download_info, reload=reload, verbose=verbose)
if self.fits_in_memory:
self.read_in_memory()
[docs] def read_in_memory(self):
"""Load full Cora data into memory and already split into items."""
filepath = os.path.join(self.data_main_dir, self.data_directory_name, "cora.npz")
loader = np.load(filepath, allow_pickle=True)
loader = dict(loader)
a = sp.csr_matrix((loader['adj_data'], loader['adj_indices'],
loader['adj_indptr']), shape=loader['adj_shape'])
x = sp.csr_matrix((loader['attr_data'], loader['attr_indices'],
loader['attr_indptr']), shape=loader['attr_shape'])
# Original adjacency matrix
# self.graph_adjacency = a
# Compute labels
labels = loader.get('labels')
labels = np.expand_dims(labels, axis=-1)
labels = np.array(labels == np.arange(70), dtype="float")
self.assign_property("node_labels", [labels]) # One graph
# Node attributes
self.assign_property("node_attributes", [x.toarray()])
# Set edges and indices.
edi, ed = convert_scaled_adjacency_to_list(a)
self.assign_property("edge_indices", [edi])
self.assign_property("edge_attributes", [np.expand_dims(ed, axis=-1)])
self.assign_property("edge_weights", [np.expand_dims(ed, axis=-1)])
# Information
self.data_keys = loader.get('idx_to_class')
return self