RecPack: An(other) Experimentation Toolkit for Top-N Recommendation using Implicit Feedback Data

Abstract

RecPack is an easy-to-use, flexible and extensible toolkit for top-N recommendation with implicit feedback data. Its goal is to support researchers with the development of their recommendation algorithms, from similarity-based to deep learning algorithms, and allow for correct, reproducible and reusable experimentation. In this demo, we give an overview of the package and show how researchers can use it to their advantage when developing recommendation algorithms.