Domain Word Sense Disambiguation

The data released on this page are related to our experiments on multi-domain acquisition from the Web and domain word sense disambiguation. Please see the following paper:

Stefano Faralli, Roberto Navigli. A New Minimally-Supervised Framework for Domain Word Sense Disambiguation. Proceedings of the 2012 Conference on Empirical Methods in Natural Language Processing (EMNLP 2012), Jeju, Korea, 2012, pp. 1411-1422.

We are releasing two packages:

Relation Seeds, which contains 31 sets of 8 domain hyponym/hypernym relation seeds we used to bootstrap the glossary acquisition phase.
Domain Sense Inventory, which contains the resulting multi-domain glossary we harvested from the Web by means of our glossary bootstrapping technique. The inventory consists of 343,320 domain senses (176,512 distinct terms and 223,247 distinct glosses).

Downloads

Relation Seeds (v1.0, revision 06 July 2012).
Domain Sense Inventory ( v1.0, revision 06 July 2012).

Each package contains a readme.txt file which describes the package content and the file formats.

References

When referring to this data, please cite the following paper:

Last update: 16 July 2012 by Stefano Faralli