Do supervised distributional methods really learn lexical inference relations?

Omer Levy; Steffen Remus; Chris Biemann; Ido Dagan

doi:10.3115/v1/n15-1098

Do supervised distributional methods really learn lexical inference relations?

Omer Levy, Steffen Remus, Chris Biemann, Ido Dagan

Department of Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

177 Scopus citations

Abstract

Distributional representations of words have been recently used in supervised settings for recognizing lexical inference relations between word pairs, such as hypernymy and entailment. We investigate a collection of these state-of-the-art methods, and show that they do not actually learn a relation between two words. Instead, they learn an independent property of a single word in the pair: whether that word is a "prototypical hypernym".

Original language	English
Title of host publication	NAACL HLT 2015 - 2015 Conference of the North American Chapter of the Association for Computational Linguistics
Subtitle of host publication	Human Language Technologies, Proceedings of the Conference
Publisher	Association for Computational Linguistics (ACL)
Pages	970-976
Number of pages	7
ISBN (Electronic)	9781941643495
DOIs	https://doi.org/10.3115/v1/n15-1098
State	Published - 2015
Event	Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL HLT 2015 - Denver, United States Duration: 31 May 2015 → 5 Jun 2015

Publication series

Name	NAACL HLT 2015 - 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference

Conference

Conference	Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL HLT 2015
Country/Territory	United States
City	Denver
Period	31/05/15 → 5/06/15

Bibliographical note

Publisher Copyright:
© 2015 Association for Computational Linguistics.

Access to Document

10.3115/v1/n15-1098

Cite this

Levy, O., Remus, S., Biemann, C., & Dagan, I. (2015). Do supervised distributional methods really learn lexical inference relations? In NAACL HLT 2015 - 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference (pp. 970-976). (NAACL HLT 2015 - 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference). Association for Computational Linguistics (ACL). https://doi.org/10.3115/v1/n15-1098

Levy, Omer ; Remus, Steffen ; Biemann, Chris et al. / Do supervised distributional methods really learn lexical inference relations?. NAACL HLT 2015 - 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference. Association for Computational Linguistics (ACL), 2015. pp. 970-976 (NAACL HLT 2015 - 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference).

@inproceedings{51b1041905444915b44af9122fb77577,

title = "Do supervised distributional methods really learn lexical inference relations?",

abstract = "Distributional representations of words have been recently used in supervised settings for recognizing lexical inference relations between word pairs, such as hypernymy and entailment. We investigate a collection of these state-of-the-art methods, and show that they do not actually learn a relation between two words. Instead, they learn an independent property of a single word in the pair: whether that word is a {"}prototypical hypernym{"}.",

author = "Omer Levy and Steffen Remus and Chris Biemann and Ido Dagan",

note = "Publisher Copyright: {\textcopyright} 2015 Association for Computational Linguistics.; Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL HLT 2015 ; Conference date: 31-05-2015 Through 05-06-2015",

year = "2015",

doi = "10.3115/v1/n15-1098",

language = "אנגלית",

series = "NAACL HLT 2015 - 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference",

publisher = "Association for Computational Linguistics (ACL)",

pages = "970--976",

booktitle = "NAACL HLT 2015 - 2015 Conference of the North American Chapter of the Association for Computational Linguistics",

address = "ארצות הברית",

}

Levy, O, Remus, S, Biemann, C & Dagan, I 2015, Do supervised distributional methods really learn lexical inference relations? in NAACL HLT 2015 - 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference. NAACL HLT 2015 - 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference, Association for Computational Linguistics (ACL), pp. 970-976, Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL HLT 2015, Denver, United States, 31/05/15. https://doi.org/10.3115/v1/n15-1098

Do supervised distributional methods really learn lexical inference relations? / Levy, Omer; Remus, Steffen; Biemann, Chris et al.
NAACL HLT 2015 - 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference. Association for Computational Linguistics (ACL), 2015. p. 970-976 (NAACL HLT 2015 - 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review