takuti/stream-recommender

Experiments of item recommendation in a streaming environment

Item Recommendation in a Streaming Environment

This repository discloses implementation used in the following research papers:

• T. Kitazawa. Incremental Factorization Machines for Persistently Cold-Starting Online Item Recommendation. arXiv:1607.02858 [cs.LG], July 2016.
• T. Kitazawa. Sketching Dynamic User-Item Interactions for Online Item Recommendation. In Proc. of CHIIR 2017, March 2017.

Recommendation algorithms are implemented in FluRS, a Python library for online item recommendation tasks.

Usage

$ python experiment.py -f path/to/config/file.ini

Examples of config files are available at: config/

The results will be written text files under results/.