Welcome to the SWANN repository! This project focuses on developing a highly efficient indexing and retrieval system for high-dimensional binary vectors using Locality-Sensitive Hashing (LSH). It incorporates trie-based indexing and efficient bucket distribution techniques to enhance search performance.
The project report provides a comprehensive study of our LSH algorithm, covering its theory, implementation, and performance analysis. It explores fundamental LSH concepts such as hash families, bucket distributions, and failure probabilities.
If you want to compile and run:
$ docker compose up
If you want to only compile write:
$ docker compose up compile
If you want to only compile and test write:
$ docker compose up test
To run benchmarking:
$ docker compose up benchmark