Directional distributional similarity for lexical inference

Lili Kotlerman; Ido Dagan; Idan Szpektor; Maayan Zhitomirsky-Geffet

doi:10.1017/s1351324910000124

Directional distributional similarity for lexical inference

Lili Kotlerman, Ido Dagan, Idan Szpektor, Maayan Zhitomirsky-Geffet

Yahoo Research Labs

Research output: Contribution to journal › Article › peer-review

173 Scopus citations

Abstract

Distributional word similarity is most commonly perceived as a symmetric relation. Yet, directional relations are abundant in lexical semantics and in many Natural Language Processing (NLP) settings that require lexical inference, making symmetric similarity measures less suitable for their identification. This paper investigates the nature of directional (asymmetric) similarity measures that aim to quantify distributional feature inclusion. We identify desired properties of such measures for lexical inference, specify a particular measure based on Average Precision that addresses these properties, and demonstrate the empirical benefit of directional measures for two different NLP datasets.

Original language	English
Pages (from-to)	359-389
Number of pages	31
Journal	Natural Language Engineering
Volume	16
Issue number	4
DOIs	https://doi.org/10.1017/s1351324910000124
State	Published - Oct 2010

Bibliographical note

Funding Information:
This work was supported by Fundação de Amparo à Pesquisa do Estado do Rio de Janeiro ( E-26/203.320/2017 ; E-26/010.101056/2018 and E-26/201.451/2014 ); Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brasil (Finance Code 001); Conselho Nacional de Desenvolvimento Científico e Tecnológico ( 305940/2018-0 ).

Funding

Funders	Funder number
Seventh Framework Programme	216886

Access to Document

10.1017/s1351324910000124

Cite this

@article{6c6a5203362c40e6b39df6529cb804cc,

title = "Directional distributional similarity for lexical inference",

abstract = "Distributional word similarity is most commonly perceived as a symmetric relation. Yet, directional relations are abundant in lexical semantics and in many Natural Language Processing (NLP) settings that require lexical inference, making symmetric similarity measures less suitable for their identification. This paper investigates the nature of directional (asymmetric) similarity measures that aim to quantify distributional feature inclusion. We identify desired properties of such measures for lexical inference, specify a particular measure based on Average Precision that addresses these properties, and demonstrate the empirical benefit of directional measures for two different NLP datasets.",

author = "Lili Kotlerman and Ido Dagan and Idan Szpektor and Maayan Zhitomirsky-Geffet",

note = "Funding Information: This work was supported by Funda{\c c}{\~a}o de Amparo {\`a} Pesquisa do Estado do Rio de Janeiro ( E-26/203.320/2017 ; E-26/010.101056/2018 and E-26/201.451/2014 ); Coordena{\c c}{\~a}o de Aperfei{\c c}oamento de Pessoal de N{\'i}vel Superior - Brasil (Finance Code 001); Conselho Nacional de Desenvolvimento Cient{\'i}fico e Tecnol{\'o}gico ( 305940/2018-0 ). ",

year = "2010",

month = oct,

doi = "10.1017/s1351324910000124",

language = "אנגלית",

volume = "16",

pages = "359--389",

journal = "Natural Language Engineering",

issn = "1351-3249",

publisher = "Cambridge University Press",

number = "4",

}

TY - JOUR

T1 - Directional distributional similarity for lexical inference

AU - Kotlerman, Lili

AU - Dagan, Ido

AU - Szpektor, Idan

AU - Zhitomirsky-Geffet, Maayan

N1 - Funding Information: This work was supported by Fundação de Amparo à Pesquisa do Estado do Rio de Janeiro ( E-26/203.320/2017 ; E-26/010.101056/2018 and E-26/201.451/2014 ); Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brasil (Finance Code 001); Conselho Nacional de Desenvolvimento Científico e Tecnológico ( 305940/2018-0 ).

PY - 2010/10

Y1 - 2010/10

N2 - Distributional word similarity is most commonly perceived as a symmetric relation. Yet, directional relations are abundant in lexical semantics and in many Natural Language Processing (NLP) settings that require lexical inference, making symmetric similarity measures less suitable for their identification. This paper investigates the nature of directional (asymmetric) similarity measures that aim to quantify distributional feature inclusion. We identify desired properties of such measures for lexical inference, specify a particular measure based on Average Precision that addresses these properties, and demonstrate the empirical benefit of directional measures for two different NLP datasets.

AB - Distributional word similarity is most commonly perceived as a symmetric relation. Yet, directional relations are abundant in lexical semantics and in many Natural Language Processing (NLP) settings that require lexical inference, making symmetric similarity measures less suitable for their identification. This paper investigates the nature of directional (asymmetric) similarity measures that aim to quantify distributional feature inclusion. We identify desired properties of such measures for lexical inference, specify a particular measure based on Average Precision that addresses these properties, and demonstrate the empirical benefit of directional measures for two different NLP datasets.

UR - http://www.scopus.com/inward/record.url?scp=78650065357&partnerID=8YFLogxK

U2 - 10.1017/s1351324910000124

DO - 10.1017/s1351324910000124

M3 - ???researchoutput.researchoutputtypes.contributiontojournal.article???

AN - SCOPUS:78650065357

SN - 1351-3249

VL - 16

SP - 359

EP - 389

JO - Natural Language Engineering

JF - Natural Language Engineering

IS - 4

ER -

Directional distributional similarity for lexical inference

Abstract

Bibliographical note

Funding

Access to Document

Other files and links

Fingerprint

Cite this