GNN Recruitment (gnnr)

Recruitment process enhanced by Graph Neural Networks (GNNs)

Prerequirements:

• Install uv

Clone the repo and change directory:

git clone https://github.com/dartt0n/gnn-recruitment.git && cd gnn-recruitment

Notebooks

Different python notebooks are available in the experiments directory.

Premade notebooks

Notebook	Colab
001-gin-shared
002-sage-shared
003-gat-shared
004-rgnn-shared
005-cgnn-shared

Manual

To run them, you need to install the dependencies. The best way to do it is to use uv.

git clone https://github.com/dartt0n/gnn-recruitment.git && cd gnn-recruitment
uv sync # install base dependencies
uv pip install torch-scatter torch-sparse -f https://data.pyg.org/whl/torch-2.5.0+cpu.html # for cpu
uv pip install torch-scatter torch-sparse -f https://data.pyg.org/whl/torch-2.5.0+cu121.html # for gpu

uv sync --extra cli # if you want to use CLI
uv sync --extra jupyter # if you want to use Jupyter
uv sync --extra cli --extra jupyter # for both

However, if you are using Google Colab, Kaggle, or InnoDataHub you can just add the following command at the top of the notebook:

1. Google Colab or Kaggle

!pip install datasets lightning wandb sentence-transformers polars torch-geometric torch-scatter torch-sparse -f https://data.pyg.org/whl/torch-2.5.0+cu121.html

1. InnoDataHub

!pip install datasets lightning wandb sentence-transformers polars torch-geometric torch-scatter torch-sparse -f https://data.pyg.org/whl/torch-2.1.1+cu119.html

Create Python environment

uv sync # install base dependencies
uv pip install torch-scatter torch-sparse -f https://data.pyg.org/whl/torch-2.5.0+cpu.html # for cpu
uv pip install torch-scatter torch-sparse -f https://data.pyg.org/whl/torch-2.5.0+cu121.html # for gpu

uv sync --extra cli # if you want to use CLI
uv sync --extra jupyter # if you want to use Jupyter
uv sync --extra cli --extra jupyter # for both

Generate data

# access help
uv run gnnr --help
uv run gnnr data --help
uv run gnnr data generate --help

# generate data with synthetic data
uv run gnnr data generate --output ./data
# or
uv run gnnr data generate --output ./data --no-synthetic

# also generate synthetic data
uv run gnnr data generate --output ./data --synthetic

# specify model to use for embeddings
uv run gnnr data generate --output ./data \
    --synthetic \
    --model sentence-transformers/all-MiniLM-L12-v2

# only synthetic data
uv run gnnr data synthetic \
    --model sentence-transformers/all-MiniLM-L12-v2 \
    --output ./data

Export dataset

Export dataset for different visualization tools

uv run gnnr data export \
    [edgelist|graphml|gexf|dot] \
    --input data/dataset/edges.json \
    --output ~/Downloads/edgelist.csv

Format code

uv run ruff format .