metapath2vec

metapath2vec: Scalable Representation Learning for Heterogeneous Networks

metapath2vec is a algorithm framework for representation learning in heterogeneous networks which contains multiple types of nodes and links. Given a heterogeneous graph, metapath2vec algorithm first generates meta-path-based random walks and then use skipgram model to train a language model. Based on PGL, we reproduce metapath2vec algorithm using PGL graph engine for scalable representation learning.

Dependencies

• paddlepaddle>=2.1.0

• pgl>=2.1.4

• OpenMPI==1.4.1

Datasets

You can download datasets from here.

We use the "aminer" data for example. After downloading the aminer data, put them, let's say, in ./data/net_aminer/. We also need to move the label/ directory to ./data/ directory.

Data preprocessing

After downloading the dataset, run the folowing command to preprocess the data:

python data_preprocess.py --config config.yaml

Hyperparameters

All the hyper parameters are saved in config.yaml file. So before training, you can open the config.yaml to modify the hyper parameters as you like.

PGL Graph Engine Launching

Now we support distributed loading graph data using PGL Graph Engine. We also develop a simple tutorial to show how to launch a graph engine, please refer to here.

To launch a distributed graph service, please follow the steps below.

IP address setting

The first step is to set the IP list for each graph server. Each IP address with port represents a server. In ip_list.txt file, we set up 4 ip addresses as follow for demo:

127.0.0.1:8553
127.0.0.1:8554
127.0.0.1:8555
127.0.0.1:8556

Launching Graph Engine by OpenMPI

Before launching the graph engine, you should set up the below hyper-parameters in config.yaml:

etype2files: "p2a:./graph_data/paper2author_edges.txt,p2c:./graph_data/paper2conf_edges.txt"
ntype2files: "p:./graph_data/node_types.txt,a:./graph_data/node_types.txt,c:./graph_data/node_types.txt"
symmetry: True
shard_num: 100

Then, we can launch the graph engine with the help of OpenMPI.

mpirun -np 4 python -m pgl.distributed.launch --ip_config ./ip_list.txt --conf ./config.yaml --mode mpi --shard_num 100

Launching Graph Engine manually

If you didn't install OpenMPI, you can launch the graph engine manually.

Fox example, if we want to use 4 servers, we should run the following command separately on 4 terminals.

# terminal 3
python -m pgl.distributed.launch --ip_config ./ip_list.txt --conf ./config.yaml --shard_num 100 --server_id 3

# terminal 2
python -m pgl.distributed.launch --ip_config ./ip_list.txt --conf ./config.yaml --shard_num 100 --server_id 2

# terminal 1
python -m pgl.distributed.launch --ip_config ./ip_list.txt --conf ./config.yaml --shard_num 100 --server_id 1

# terminal 0
python -m pgl.distributed.launch --ip_config ./ip_list.txt --conf ./config.yaml --shard_num 100 --server_id 0

Note that the server_id of 0 should be the last one to be launched.

Training

After successfully launching the graph engine, you can run the below command to train the model.

export CUDA_VISIBLE_DEVICES=0
python train.py --config ./config.yaml --ip ./ip_list.txt

Note that the trained model will be saved ./ckpt_custom/$task_name/