Recommendation Systems: A Reinforcement Learning Approach of the Cold-User Problem

Everything is packaged in a Conda environment. You can create the environment with all the necessary packages using the environment.yml file. To create the environment, run:

conda env create -f environment.yml

Make sure you have conda installed on your machine.

The Active Learning Strategies are in the active_learning.py or by using jupyter notebook in active_learning_notebook.ipynb. To run, either use Jupyter Notebook to run the cells or use the following command:

python active_learning.py

Make sure you're using the conda virtual environment and the dataset is present.

The Reinforcement Learning approach code is in the relecur.py or by using jupyter notebook in relecur_notebook.ipynb. Again, either use Jupyter Notebook to run the cells or use the following command:

python relecur.py

The Reinforcement Learning environments are in the environment_al.py and in environment_items.py files. The former contains the environment for the AL-based method, and the latter contains the environment for the Item-based method. These environment are imported in the relecur.py file.