Simulated Contextual Bandits from Recommendation Datasets

Source code for the numerical experiments presented in the paper "Simulated Contextual Bandits for Personalization Tasks from Recommendation Datasets", preprint is available at https://arxiv.org/abs/2210.10631

Setup

• Install the requirements via pip install -r requirements.txt
• Run the code via python -m main -e {evn} where {env} is one of the recommendation datasets, i.e. e = {ml25m, imdb}