Skip to content

Latest commit

 

History

History

notebooks

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
Baselines.ipynb
Baselines.ipynb
 
 
ConvertingToGraphs.ipynb
ConvertingToGraphs.ipynb
 
 
Embeddings.ipynb
Embeddings.ipynb
 
 
Filtering.ipynb
Filtering.ipynb
 
 
README.md
README.md
 
 

README.md

Java Local Import Prediction with Gated Graph Neural Networks

Data Preprocessing and Baselines

This directory contains several ipython notebooks that can be used to preprocess the data downloaded from BigQuery and parsed with JavaParser. You might need the following packages to run this code: pytorch, numpy, tqdm, joblib, networkx, sklearn, matplotlib, pandas. Below is the breakdown of what each of the notebooks is for:

  • Filtering.ipynb - contains code that can be used to filter out repositories that are duplicates of other repositories, repositories that contain files that could not be parsed, repositories that contain files that do not conform with java package naming conventions, etc.

  • Embeddings.ipynb - defines a simple autoencoder that can be used to reduce the dimensionality of glove embeddings so that they can be successfully used to create filename embeddings in ConvertingToGraphs.ipynb

  • ConvertingToGraphs.ipynb - contains code that allows to convert data generated by Filterings.ipynb to a format that could serve as input to a GGNN. Also shows some statistics and examples of how such graphs might look like

  • Baselines.ipynb - defines a set of relatively simple baselines that the GGNN can be compared to in terms of performance on the input prediction task. These include embedding similarity, node degree, shortest path from one node to another in a graph, etc.