Knowledge Augmented Abstractive Summarizer

Overview

This repository contains the implementation of a Machine Learning project for the Machine Learning for Humanities course. The project focuses on knowledge-augmented summarization, leveraging external data sources to enhance text generation quality and factual consistency.

Some examples of knowledge augmented summaries can be found in the dataset_comparisons.txt and comparison.txt files.

Setup

To set up and run this project, follow these steps:

Set Up Environment Variables The project requires an API key for the Mistral model and an email for Wikipedia API access. Create a .env file in the folder containing the API integrations and add the following:

MISTRAL_API_KEY=your_mistral_api_key
WIKI_ACC=your_wikipedia_email

Dowload Packages

pip install spacy
python -m spacy download en_core_web_sm
pip install requests
pip install SPARQLWrapper
pip install transformers
pip install torch
pip install rouge-score
pip install datasets
pip install bert-score
pip install fuzzywuzzy
pip install python-dotenv

License

This project is licensed under the MIT License.