A Tensorflow Implementation of Gated Graph Neural Networks (GGNN) for Method Name Prediction

Download data

python3 download_data.py

This script will download the three datasets java-small, java-med and java-large (only the testing part) to the directory data/

Download pretrained model

python3 download_pretrained_model.py

This script will download the pretrained models for the three datasets into the directory model/

Test the model

source test_ggnn_method_name_prediction.sh

It will print some thing like this on the console:

Then it means that you have done the steps correctly.

There is a parameter called Dataset in the script test_ggnn_method_name_prediction.sh, it will point to the corresponding model and the corresponding dataset to test. At the moment, default is Dataset=java-small, you can change to Dataset=java-med or Dataset=java-large for test on the java-med and java-large dataset.