easygraph.datasets package

Subpackages

Submodules

easygraph.datasets.citation_graph module

Cora, citeseer, pubmed dataset.

(lingfan): following dataset loading and preprocessing code from tkipf/gcn https://github.com/tkipf/gcn/blob/master/gcn/utils.py

class easygraph.datasets.citation_graph.CitationGraphDataset(name, raw_dir=None, force_reload=False, verbose=True, reverse_edge=True, transform=None, reorder=False)[source]

Bases: EasyGraphBuiltinDataset

The citation graph dataset, including cora, citeseer and pubmeb. Nodes mean authors and edges mean citation relationships.

Parameters:

  • name (str) – name can be ‘cora’, ‘citeseer’ or ‘pubmed’.

  • raw_dir (str) – Raw file directory to download/contains the input data directory. Default: ~/.dgl/

  • force_reload (bool) – Whether to reload the dataset. Default: False

  • verbose (bool) – Whether to print out progress information. Default: True.

  • reverse_edge (bool) – Whether to add reverse edges in graph. Default: True.

  • transform (callable, optional) – A transform that takes in a DGLGraph object and returns a transformed version. The DGLGraph object will be transformed before every access.

  • reorder (bool) – Whether to reorder the graph using reorder_graph(). Default: False.

has_cache()[source]

Overwrite to realize your own logic of deciding whether there exists a cached dataset.

By default False.

property num_classes
property num_labels
process()[source]

Loads input data from data directory and reorder graph for better locality

ind.name.x => the feature vectors of the training instances as scipy.sparse.csr.csr_matrix object; ind.name.tx => the feature vectors of the test instances as scipy.sparse.csr.csr_matrix object; ind.name.allx => the feature vectors of both labeled and unlabeled training instances

(a superset of ind.name.x) as scipy.sparse.csr.csr_matrix object;

ind.name.y => the one-hot labels of the labeled training instances as numpy.ndarray object; ind.name.ty => the one-hot labels of the test instances as numpy.ndarray object; ind.name.ally => the labels for instances in ind.name.allx as numpy.ndarray object; ind.name.graph => a dict in the format {index: [index_of_neighbor_nodes]} as collections.defaultdict

object;

ind.name.test.index => the indices of test instances in graph, for the inductive setting as list object.

property reverse_edge
property save_name
class easygraph.datasets.citation_graph.CiteseerGraphDataset(raw_dir=None, force_reload=False, verbose=True, reverse_edge=True, transform=None, reorder=False)[source]

Bases: CitationGraphDataset

Citeseer citation network dataset.

Nodes mean scientific publications and edges mean citation relationships. Each node has a predefined feature with 3703 dimensions. The dataset is designed for the node classification task. The task is to predict the category of certain publication.

Statistics:

  • Nodes: 3327

  • Edges: 9228

  • Number of Classes: 6

  • Label Split:

    • Train: 120

    • Valid: 500

    • Test: 1000

Parameters:

  • raw_dir (str) – Raw file directory to download/contains the input data directory. Default: ~/.dgl/

  • force_reload (bool) – Whether to reload the dataset. Default: False

  • verbose (bool) – Whether to print out progress information. Default: True.

  • reverse_edge (bool) – Whether to add reverse edges in graph. Default: True.

  • transform (callable, optional) – A transform that takes in a DGLGraph object and returns a transformed version. The DGLGraph object will be transformed before every access.

  • reorder (bool) – Whether to reorder the graph using reorder_graph(). Default: False.

num_classes

Number of label classes

Type:

int

Notes

The node feature is row-normalized.

In citeseer dataset, there are some isolated nodes in the graph. These isolated nodes are added as zero-vecs into the right position.

Examples

>>> dataset = CiteseerGraphDataset()
>>> g = dataset[0]
>>> num_class = dataset.num_classes
>>>
>>> # get node feature
>>> feat = g.ndata['feat']
>>>
>>> # get data split
>>> train_mask = g.ndata['train_mask']
>>> val_mask = g.ndata['val_mask']
>>> test_mask = g.ndata['test_mask']
>>>
>>> # get labels
>>> label = g.ndata['label']
class easygraph.datasets.citation_graph.CoraBinary(raw_dir=None, force_reload=False, verbose=True, transform=None)[source]

Bases: EasyGraphBuiltinDataset

A mini-dataset for binary classification task using Cora.

After loaded, it has following members:

graphs : list of DGLGraph pmpds : list of scipy.sparse.coo_matrix labels : list of numpy.ndarray

Parameters:

  • raw_dir (str) – Raw file directory to download/contains the input data directory. Default: ~/.dgl/

  • force_reload (bool) – Whether to reload the dataset. Default: False

  • verbose (bool) – Whether to print out progress information. Default: True.

  • transform (callable, optional) – A transform that takes in a DGLGraph object and returns a transformed version. The DGLGraph object will be transformed before every access.

has_cache()[source]

Overwrite to realize your own logic of deciding whether there exists a cached dataset.

By default False.

process()[source]

Overwrite to realize your own logic of processing the input data.

property save_name
class easygraph.datasets.citation_graph.CoraGraphDataset(raw_dir=None, force_reload=False, verbose=True, reverse_edge=True, transform=None, reorder=False)[source]

Bases: CitationGraphDataset

Cora citation network dataset.

Nodes mean paper and edges mean citation relationships. Each node has a predefined feature with 1433 dimensions. The dataset is designed for the node classification task. The task is to predict the category of certain paper.

Statistics:

  • Nodes: 2708

  • Edges: 10556

  • Number of Classes: 7

  • Label split:

    • Train: 140

    • Valid: 500

    • Test: 1000

Parameters:

  • raw_dir (str) – Raw file directory to download/contains the input data directory. Default: ~/.dgl/

  • force_reload (bool) – Whether to reload the dataset. Default: False

  • verbose (bool) – Whether to print out progress information. Default: True.

  • reverse_edge (bool) – Whether to add reverse edges in graph. Default: True.

  • transform (callable, optional) – A transform that takes in a DGLGraph object and returns a transformed version. The DGLGraph object will be transformed before every access.

  • reorder (bool) – Whether to reorder the graph using reorder_graph(). Default: False.

num_classes

Number of label classes

Type:

int

Notes

The node feature is row-normalized.

Examples

>>> dataset = CoraGraphDataset()
>>> g = dataset[0]
>>> num_class = dataset.num_classes
>>>
>>> # get node feature
>>> feat = g.ndata['feat']
>>>
>>> # get data split
>>> train_mask = g.ndata['train_mask']
>>> val_mask = g.ndata['val_mask']
>>> test_mask = g.ndata['test_mask']
>>>
>>> # get labels
>>> label = g.ndata['label']
class easygraph.datasets.citation_graph.PubmedGraphDataset(raw_dir=None, force_reload=False, verbose=True, reverse_edge=True, transform=None, reorder=False)[source]

Bases: CitationGraphDataset

Pubmed citation network dataset.

Nodes mean scientific publications and edges mean citation relationships. Each node has a predefined feature with 500 dimensions. The dataset is designed for the node classification task. The task is to predict the category of certain publication.

Statistics:

  • Nodes: 19717

  • Edges: 88651

  • Number of Classes: 3

  • Label Split:

    • Train: 60

    • Valid: 500

    • Test: 1000

Parameters:

  • raw_dir (str) – Raw file directory to download/contains the input data directory. Default: ~/.dgl/

  • force_reload (bool) – Whether to reload the dataset. Default: False

  • verbose (bool) – Whether to print out progress information. Default: True.

  • reverse_edge (bool) – Whether to add reverse edges in graph. Default: True.

  • transform (callable, optional) – A transform that takes in a DGLGraph object and returns a transformed version. The DGLGraph object will be transformed before every access.

  • reorder (bool) – Whether to reorder the graph using reorder_graph(). Default: False.

num_classes

Number of label classes

Type:

int

Notes

The node feature is row-normalized.

Examples

>>> dataset = PubmedGraphDataset()
>>> g = dataset[0]
>>> num_class = dataset.num_of_class
>>>
>>> # get node feature
>>> feat = g.ndata['feat']
>>>
>>> # get data split
>>> train_mask = g.ndata['train_mask']
>>> val_mask = g.ndata['val_mask']
>>> test_mask = g.ndata['test_mask']
>>>
>>> # get labels
>>> label = g.ndata['label']
easygraph.datasets.citation_graph.load_citeseer(raw_dir=None, force_reload=False, verbose=True, reverse_edge=True, transform=None)[source]

Get CiteseerGraphDataset

Parameters:

  • raw_dir (str) – Raw file directory to download/contains the input data directory. Default: ~/.dgl/

  • force_reload (bool) – Whether to reload the dataset. Default: False

  • verbose (bool) – Whether to print out progress information. Default: True.

  • reverse_edge (bool) – Whether to add reverse edges in graph. Default: True.

  • transform (callable, optional) – A transform that takes in a DGLGraph object and returns a transformed version. The DGLGraph object will be transformed before every access.

Return type:

CiteseerGraphDataset

easygraph.datasets.citation_graph.load_cora(raw_dir=None, force_reload=False, verbose=True, reverse_edge=True, transform=None)[source]

Get CoraGraphDataset

Parameters:

  • raw_dir (str) – Raw file directory to download/contains the input data directory. Default: ~/.dgl/

  • force_reload (bool) – Whether to reload the dataset. Default: False

  • verbose (bool) – Whether to print out progress information. Default: True.

  • reverse_edge (bool) – Whether to add reverse edges in graph. Default: True.

  • transform (callable, optional) – A transform that takes in a DGLGraph object and returns a transformed version. The DGLGraph object will be transformed before every access.

Return type:

CoraGraphDataset

easygraph.datasets.citation_graph.load_pubmed(raw_dir=None, force_reload=False, verbose=True, reverse_edge=True, transform=None)[source]

Get PubmedGraphDataset

Parameters:

  • raw_dir (str) – Raw file directory to download/contains the input data directory. Default: ~/.dgl/

  • force_reload (bool) – Whether to reload the dataset. Default: False

  • verbose (bool) – Whether to print out progress information. Default: True.

  • reverse_edge (bool) – Whether to add reverse edges in graph. Default: True.

  • transform (callable, optional) – A transform that takes in a DGLGraph object and returns a transformed version. The DGLGraph object will be transformed before every access.

Return type:

PubmedGraphDataset

easygraph.datasets.get_sample_graph module

easygraph.datasets.get_sample_graph.get_graph_blogcatalog()[source]

Returns the undirected graph of blogcatalog.

Returns:

get_graph_blogcatalog – The undirected graph instance of blogcatalog from dataset: https://github.com/phanein/deepwalk/blob/master/example_graphs/blogcatalog.mat

Return type:

easygraph.Graph

References

easygraph.datasets.get_sample_graph.get_graph_flickr()[source]

Returns the undirected graph of Flickr dataset.

Returns:

get_graph_flickr – The undirected graph instance of Flickr from dataset: http://socialnetworks.mpi-sws.mpg.de/data/flickr-links.txt.gz

Return type:

easygraph.Graph

References

easygraph.datasets.get_sample_graph.get_graph_karateclub()[source]

Returns the undirected graph of Karate Club.

Returns:

get_graph_karateclub – The undirected graph instance of karate club from dataset: http://vlado.fmf.uni-lj.si/pub/networks/data/Ucinet/UciData.htm

Return type:

easygraph.Graph

References

easygraph.datasets.get_sample_graph.get_graph_youtube()[source]

Returns the undirected graph of Youtube dataset.

Returns:

get_graph_youtube – The undirected graph instance of Youtube from dataset: http://socialnetworks.mpi-sws.mpg.de/data/youtube-links.txt.gz

Return type:

easygraph.Graph

References

easygraph.datasets.gnn_benchmark module

class easygraph.datasets.gnn_benchmark.AmazonCoBuyComputerDataset(raw_dir=None, force_reload=False, verbose=True, transform=None)[source]

Bases: GNNBenchmarkDataset

‘Computer’ part of the AmazonCoBuy dataset for node classification task.

Amazon Computers and Amazon Photo are segments of the Amazon co-purchase graph [McAuley et al., 2015], where nodes represent goods, edges indicate that two goods are frequently bought together, node features are bag-of-words encoded product reviews, and class labels are given by the product category.

Reference: https://github.com/shchur/gnn-benchmark#datasets

Statistics:

  • Nodes: 13,752

  • Edges: 491,722 (note that the original dataset has 245,778 edges but DGL adds the reverse edges and remove the duplicates, hence with a different number)

  • Number of classes: 10

  • Node feature size: 767

Parameters:

  • raw_dir (str) – Raw file directory to download/contains the input data directory. Default: ~/.dgl/

  • force_reload (bool) – Whether to reload the dataset. Default: False

  • verbose (bool) – Whether to print out progress information. Default: True.

  • transform (callable, optional) – A transform that takes in a DGLGraph object and returns a transformed version. The DGLGraph object will be transformed before every access.

num_classes

Number of classes for each node.

Type:

int

Examples

>>> data = AmazonCoBuyComputerDataset()
>>> g = data[0]
>>> num_class = data.num_classes
>>> feat = g.ndata['feat']  # get node feature
>>> label = g.ndata['label']  # get node labels
property num_classes

Number of classes.

Return type:

int

easygraph.datasets.graph_dataset_base module

easygraph.datasets.karate module

class easygraph.datasets.karate.KarateClubDataset(transform=None)[source]

Bases: EasyGraphDataset

Karate Club dataset for Node Classification

Zachary’s karate club is a social network of a university karate club, described in the paper “An Information Flow Model for Conflict and Fission in Small Groups” by Wayne W. Zachary. The network became a popular example of community structure in networks after its use by Michelle Girvan and Mark Newman in 2002. Official website: http://konect.cc/networks/ucidata-zachary/

Karate Club dataset statistics:

  • Nodes: 34

  • Edges: 156

  • Number of Classes: 2

Parameters:

transform (callable, optional) – A transform that takes in a DGLGraph object and returns a transformed version. The DGLGraph object will be transformed before every access.

num_classes

Number of node classes

Type:

int

Examples

>>> dataset = KarateClubDataset()
>>> num_classes = dataset.num_classes
>>> g = dataset[0]
>>> labels = g.ndata['label']
property num_classes

Number of classes.

process()[source]

Overwrite to realize your own logic of processing the input data.

easygraph.datasets.ppi module

PPIDataset for inductive learning.

class easygraph.datasets.ppi.LegacyPPIDataset(mode='train', raw_dir=None, force_reload=False, verbose=False, transform=None)[source]

Bases: PPIDataset

Legacy version of PPI Dataset

class easygraph.datasets.ppi.PPIDataset(mode='train', raw_dir=None, force_reload=False, verbose=False, transform=None)[source]

Bases: EasyGraphBuiltinDataset

Protein-Protein Interaction dataset for inductive node classification

A toy Protein-Protein Interaction network dataset. The dataset contains 24 graphs. The average number of nodes per graph is 2372. Each node has 50 features and 121 labels. 20 graphs for training, 2 for validation and 2 for testing.

Reference: http://snap.stanford.edu/graphsage/

Statistics:

  • Train examples: 20

  • Valid examples: 2

  • Test examples: 2

Parameters:

  • mode (str) – Must be one of (‘train’, ‘valid’, ‘test’). Default: ‘train’

  • raw_dir (str) – Raw file directory to download/contains the input data directory. Default: ~/.dgl/

  • force_reload (bool) – Whether to reload the dataset. Default: False

  • verbose (bool) – Whether to print out progress information. Default: True.

  • transform (callable, optional) – A transform that takes in a DGLGraph object and returns a transformed version. The DGLGraph object will be transformed before every access.

num_labels

Number of labels for each node

Type:

int

labels

Node labels

Type:

Tensor

features

Node features

Type:

Tensor

Examples

>>> dataset = PPIDataset(mode='valid')
>>> num_labels = dataset.num_labels
>>> for g in dataset:
....    feat = g.ndata['feat']
....    label = g.ndata['label']
....    # your code here
>>>
has_cache()[source]

Overwrite to realize your own logic of deciding whether there exists a cached dataset.

By default False.

property num_labels
process()[source]

Overwrite to realize your own logic of processing the input data.

save()[source]

Overwrite to realize your own logic of saving the processed dataset into files.

It is recommended to use dgl.data.utils.save_graphs to save dgl graph into files and use dgl.data.utils.save_info to save extra information into files.

easygraph.datasets.utils module

easygraph.datasets.utils.download(url, path=None, overwrite=True, sha1_hash=None, retries=5, verify_ssl=True, log=True)[source]

Download a given URL.

Codes borrowed from mxnet/gluon/utils.py

Parameters:

  • url (str) – URL to download.

  • path (str, optional) – Destination path to store downloaded file. By default stores to the current directory with the same name as in url.

  • overwrite (bool, optional) – Whether to overwrite the destination file if it already exists. By default always overwrites the downloaded file.

  • sha1_hash (str, optional) – Expected sha1 hash in hexadecimal digits. Will ignore existing file when hash is specified but doesn’t match.

  • retries (integer, default 5) – The number of times to attempt downloading in case of failure or non 200 return codes.

  • verify_ssl (bool, default True) – Verify SSL certificates.

  • log (bool, default True) – Whether to print the progress for download

Returns:

The file path of the downloaded file.

Return type:

str

easygraph.datasets.utils.extract_archive(file, target_dir, overwrite=False)[source]

Extract archive file.

Parameters:

  • file (str) – Absolute path of the archive file.

  • target_dir (str) – Target directory of the archive to be uncompressed.

  • overwrite (bool, default True) – Whether to overwrite the contents inside the directory. By default always overwrites.

easygraph.datasets.utils.generate_mask_tensor(mask)[source]

Generate mask tensor according to different backend For torch, it will create a bool tensor :param mask: input mask tensor :type mask: numpy ndarray

easygraph.datasets.utils.get_download_dir()[source]

Get the absolute path to the download directory.

Returns:

dirname – Path to the download directory

Return type:

str

easygraph.datasets.utils.makedirs(path)[source]

Module contents