CLANE

Content- and Link-Aware Node Embedding in Graphs

Install

pip install git+https://github.com/helloybz/CLANE.git

Usage

clane embedding --data_root /path/to/data_root --output_root /path/to/output/root --config_file /path/to/config.yml

--data_root is supposed to be a directory which contains at least 2 files, V and E.
V contains list of node ids separated by \n.
E contains list of pairs of node id, separated by \n.
The pair of node should has a form like [source_node_id]\t[target_node_id].

Example

data_root/V

1
2
3
4
5

data_root/E

1\t2
1\t4
2\t5
3\t1

Config file

All of the hyper-parameters of deepwalk is controlled in this yaml file.
Below is an example of the config file.

graph:
  embedding_dim: 2

similarity:
  method: "CosineSimilarity"
  kwargs:
    foo: "bar"

embedder:
  gamma: 0.76
  tolerence: 10

Experiments

CORA dataset, a citation network, is used.

• Bag-of-words encoding for each node in 1,433 dimension.
• 7 classes

Multi-Label Classification

• Compare scores of classification for CLANE-applied embeddings to those of pure Bag-of-words encodings.
• Logistic regression is used as classifier.
• Train:Test ratio varies from 1:9 to 9:1.
• Metrics are averaged after 10 runs.

Micro F1

Column name means the percentage of train split.

Method	10%	20%	30%	40%	50%	60%	70%	80%	90%
CLANE-applied	0.768	0.817	0.833	0.834	0.855	0.865	0.857	0.847	0.845
Pure Bag-of-words	0.747	0.792	0.815	0.811	0.837	0.855	0.846	0.832	0.846

Macro F1

Column name means the percentage of train split.

Method	10%	20%	30%	40%	50%	60%	70%	80%	90%
Deepwalk-Clone	0.643	0.705	0.730	0.732	0.741	0.768	0.769	0.768	0.745
Deepwalk	0.607	0.672	0.708	0.708	0.714	0.748	0.755	0.750	0.720