BM25 with pseudo-relevance feedback

This repository contains the code for my final project of "Information Retrieval".

Installation

Install the required Python libraries

pip install -qr requirements.txt

Download NLTK corpora (the reuters corpus is needed only to run the demo)

import nltk
nltk.download('punkt')
nltk.download("wordnet")
nltk.download("stopwords")
nltk.download('omw-1.4')
nltk.download('reuters')

Usage

After installing the required libraries, you can run the demo:

python demo.py

Author

• Alessandro Pierro

License

This project is licensed under the MIT license.

References

[1] Stephen Robertson and Hugo Zaragoza - The Probabilistic Relevance Framework: BM25 and Beyond, Foundations and Trends in Information Retrieval, 2009

[1] Christopher D. Manning, Prabhakar Raghavan and Hinrich Schütze - Introduction to Information Retrieval, Cambridge University Press. 2008