This code performs inference, learning, and outputs a taxonomy, as described by the following paper:

Jonathan Bragg, Mausam, and Daniel S. Weld. 2013. Crowdsourcing multi-label classification for taxonomy creation. In Proceedings of the First AAAI Conference on Human Computation and Crowdsourcing (HCOMP '13). Palm Springs, CA, USA.

Install

Install Miniconda and execute the following command to install dependencies:

conda env create -f environment.yml -n deluge

Usage

• The main script run.py accepts a data file as described in preprocess.py and uses inference methods to generate taxonomies (which are printed to stdout).
• control.py and util.py are provided as references for the infogain criterion, but adaptive item selection is not fully integrated in this release.