Matrix factorization models enhanced with social tags

Implementation of matrix factorization models enhanced with social tags for the task of music recommendation, as presented in the following papers:

• Andreu Vall, Marcin Skowron, Peter Knees, and Markus Schedl. “Improving Music Recommendations with a Weighted Factorization of the Tagging Activity.” In Proc. ISMIR. Málaga, Spain, 2015.

• Andreu Vall. “Listener-Inspired Automated Music Playlist Generation.” In Proc. RecSys. Vienna, Austria, 2015.

Models implemented

This project implements the following models:

• MF: Matrix factorization for implicit feedback as in Yifan Hu, Yehuda Koren and Chris Volinsky, “Collaborative Filtering for Implicit Feedback Datasets.” in Proc. ICDM 2008.

• TMF: Matrix factorization enhanced with social tags. Greatly based on Yi Fang and Luo Si, “Matrix Co-Factorization for Recommendation with Rich Side Information and Implicit Feedback.” in Proc. HETREC 2011, but modified in our paper to specifically deal with tags.

• WTMF: Matrix factorization enhanced with weighted social tags as introduced in our papers (see above).

Usage

Each model is implemented in a separated python file. You can simply run a model, say WTMF, by doing:

python WTMF.py

This will apply the model on the dataset specified within the WTMF.py file. Note that a dummy dataset is already included in the project. You can also specifiy the dataset manually, but please note the data folder structure:

python WTMF.py <collection-name> <dataset-name>

Details

Specific parameters for the model need to be provided within each model file. A detailed description of the parameters can be found in the papers.

Each program outputs a json file with the expected percentile rank achieved by the model using the given dataset under the given parameters. A detailed description of the evaluation methodology can be found in the papers.

License

The contents of this repository are licensed. See the LICENSE file for further details.