Relation Extraction

Prerequisites

• Python 3.5.2+

• (Optional, but highly recommended) Create and use a virtual environment for isolating from system site directories - there are many options, but the most recommended is venv by the time of writing this. Creating virtual environment with venv: python3 -m venv <path_to_your_virtual_env>. Activating it is done with source <path_to_your_virtual_env>/bin/activate.

• To install the required packages, activate the virtual environment and run pip3 install -r requirements.txt or if you are without a virtual environment, run pip3 install --user -r requirements.txt to install them in a Python user install directory instead of a system directory.

Structure

data/: Contains data sets for training, testing, validation and external resources.

documentation/: Documentation of the project.

main.py: The main configurable pipeline for training, testing and running models on data sets.

models/: Contains all our models which have a fit() and predict() method.

notebooks/: Contains Jupyter notebooks.

predictions/: Contains predictions.

README.md: Documentation of the repository.

requirements.txt: List of Python pip dependencies.

results/: Contains JSON files with evaluation results of the trained models and their parameters.

transformers/: Contains all feature transformers.

utils.py: A set of utility methods used for model evaluation, and others.

Data

The data being used is the NYT10 data set found here. Download the resources and follow the instructions to set up the data set.