Read Me

This repository corresponds to the paper "Context Tree for Adaptive Session-based Recommendation" at https://arxiv.org/abs/1806.03733

Requirements

• Anaconda 4.X (Python 3.5+)
• NumPy
• SciPy
• BLAS
• Pandas
• Theano
• Sklearn

Usage

Datasets

• Two sample datasets are provided with anonymization in data folder
• Two RecSys Challenge dataset can be downloaded at https://www.dropbox.com/sh/2jw72n03t6zyde9/AAD25moSH9qmiIHe1YGewUYea?dl=0
• All datasets in the same format can be used for static configuration (run_test.py)
• full 'course3' and 'news2' datasets ordered by time are used for adpative configuration (run_test_adapt.py);

Running experiments

1. Double check datasets in data folder, or download from the link above
2. static configuration: run_test.py: train model on trainig data and evaluates all testing views per 'session'. results and computation times will be output to both terminal and a file.
3. adaptive configuration: run_test_adapt.py evaluates all views per 'event' on full dataset (ordered by time) results will be output to both terminal and a file