TaxoLearn

TaxoLearn is a graph-based approach aimed at learning a lexical taxonomy automatically starting from a domain corpus and the Web. The system is based on Word-Class Lattices and a taxonomy learning algorithm developed by Roberto Navigli, Paola Velardi and Stefano Faralli.

Clear here for an extended version of this work, called OntoLearn Reloaded.

Dataset

We are releasing the Artificial Intelligence taxonomy (extracted from the IJCAI 2009 conference proceedings). The taxonomy is distributed in two formats:

TSV - tab-separated values, consisting of three columns (HYPERNYM, ->, TERM);
DOT - a DOT graph (see the graphviz website for more information).

Downloads

We are also releasing the glosses extracted from the domain corpus and harvested from the web during the taxonomy learning phase:

Glosses (TSV)

References

If you use our IJCAI taxonomy in your own work or publish new work on the topic, please cite the following paper:

Roberto Navigli, Paola Velardi, Stefano Faralli. A Graph-based Algorithm for Inducing Lexical Taxonomies from Scratch. Proc. of the 22nd International Joint Conference on Artificial Intelligence (IJCAI 2011), Barcelona, Spain, July 19-22nd, 2011.

Last update: 10 Dec 2012 by Stefano Faralli