Hearst Patterns Revisited: Automatic Hypernym Detection from Large Text
Corpora
S. Roller, D. Kiela, and M. Nickel. ACL (2), page 358-363. Association for Computational Linguistics, (2018)
Abstract
Methods for unsupervised hypernym detection may broadly be categorized
according to two paradigms: pattern-based and distributional methods. In this
paper, we study the performance of both approaches on several hypernymy tasks
and find that simple pattern-based methods consistently outperform
distributional methods on common benchmark datasets. Our results show that
pattern-based models provide important contextual constraints which are not yet
captured in distributional methods.
%0 Conference Paper
%1 roller2018hearst
%A Roller, Stephen
%A Kiela, Douwe
%A Nickel, Maximilian
%B ACL (2)
%D 2018
%I Association for Computational Linguistics
%K embeddings hearst hierachical relations
%P 358-363
%T Hearst Patterns Revisited: Automatic Hypernym Detection from Large Text
Corpora
%U http://arxiv.org/abs/1806.03191
%X Methods for unsupervised hypernym detection may broadly be categorized
according to two paradigms: pattern-based and distributional methods. In this
paper, we study the performance of both approaches on several hypernymy tasks
and find that simple pattern-based methods consistently outperform
distributional methods on common benchmark datasets. Our results show that
pattern-based models provide important contextual constraints which are not yet
captured in distributional methods.
@inproceedings{roller2018hearst,
abstract = {Methods for unsupervised hypernym detection may broadly be categorized
according to two paradigms: pattern-based and distributional methods. In this
paper, we study the performance of both approaches on several hypernymy tasks
and find that simple pattern-based methods consistently outperform
distributional methods on common benchmark datasets. Our results show that
pattern-based models provide important contextual constraints which are not yet
captured in distributional methods.},
added-at = {2018-10-18T21:14:47.000+0200},
author = {Roller, Stephen and Kiela, Douwe and Nickel, Maximilian},
biburl = {https://www.bibsonomy.org/bibtex/25ea3c9b7c2187f477c9c8056b906c5fb/thoni},
booktitle = {ACL (2)},
interhash = {ed7de4b70108ce9124ef778e5c95539e},
intrahash = {5ea3c9b7c2187f477c9c8056b906c5fb},
keywords = {embeddings hearst hierachical relations},
pages = {358-363},
publisher = {Association for Computational Linguistics},
timestamp = {2018-10-25T01:25:29.000+0200},
title = {Hearst Patterns Revisited: Automatic Hypernym Detection from Large Text
Corpora},
url = {http://arxiv.org/abs/1806.03191},
year = 2018
}