Librec-auto

https://coveralls.io/repos/github/that-recsys-lab/librec-auto/badge.svg?branch=master

About

librec-auto is a Python tool for running recommender systems experiments. It is built on top of the open-source LibRec package, and can take advantage of the many algorithm and metric implementations there.

The basic element of librec-auto execution is the “study”, which is a series of experiments carried with a single algorithm, a single data set, and a set of evaluation metrics. The experiments differ from each other by the hyperparameters given to the algorithm. librec-auto allows such studies to be conducted with minimal experimenter intervention, and supports such capabilities as:

  • recommendation result re-ranking
  • re-running evaluations without re-computing results
  • summarizing results in output graphs
  • integration with Slack and Dropbox

More complete documentation is available at readthedocs:

Workflow

The workflow of an study involves identifying appropriate data, creating training / test splits, implementing or choosing algorithms, running experiments (possibly with a range of different parameters), and reporting on the results.

Configuration

Librec-auto uses an XML-based configuration system similar to Maven or Ant.

Project structure

This directory contains the Python libraries for the librec_auto module. There are two other affiliated respositories:

  • librec-auto-java: Contains the java source for the wrapper between LibRec and librec-auto, which is implemented in the auto.jar file.
  • librec-auto-sample: Contains sample data and configuration files that can be used to explore the functionality of librec-auto

Repo structure

  • /jar: Contains the jar files for LibRec and the wrapper.
  • /rules: Contains the rules for translating configuration data to LibRec properties format.
  • /librec_auto: Contains the Python code for the project.
  • /doc: Contains documentation for the project
  • /test: Contains the unit tests (not many right now)

Indices and tables