Difference between revisions of "2017 AnOverviewofNaturalLanguageInfe"

From GM-RKB
Jump to: navigation, search
(ContinuousReplacement)
(Tag: continuous replacement)
(Introduction)
(2 intermediate revisions by 2 users not shown)
Line 14: Line 14:
  
 
[[Understanding]] a [[Natural Language (NL) sentence]] amounts to understanding its [[consequence]]s as well as the [[sentence]]s that it is a consequence of. </s>
 
[[Understanding]] a [[Natural Language (NL) sentence]] amounts to understanding its [[consequence]]s as well as the [[sentence]]s that it is a consequence of. </s>
As [[Cooper et al. (1996]]) aptly put it ‘inferential ability is not only a central manifestation of [[semantic competence]] but is in [[fact centrally constitutive of it’]]. </s>
+
As [[Cooper et al. (1996]]) aptly put it "[[inferential ability]] is not only a central manifestation of [[semantic competence]] but is in [[fact]] centrally constitutive of it]]". </s>
Given the [[latter claim]], it is no surprise that the same [[author]]s argue that the best way to [[test]] the [[semantic adequacy of NLP system]]s is to look at how well they perform with regard to [[Natural Language Inference (NLI)]]. </s>
+
Given the [[latter claim]], it is no surprise that the same [[author]]s argue that the best way to [[test]] the [[semantic adequacy]] of [[NLP system]]s is to look at how well they perform with regard to [[Natural Language Inference (NLI)]]. </s>
It is [[very hard not to agree]] with [[this statement]]. </s>
+
It is very hard not to agree with [[this statement]]. </s>
 
Indeed, [[NLI]] is considered by many [[researcher]]s to be the crux of [[computational semantic]]s. </s>
 
Indeed, [[NLI]] is considered by many [[researcher]]s to be the crux of [[computational semantic]]s. </s>
[[This paper is about the datasets created]] for [[this need]]. </s>
+
[[This paper]] is about the [[dataset]]s created for this need. </s>
In particular, [[we]] discuss the most common [[NLI resource]]s arguing that all these a) fail to capture the wealth of [[inferential mechanism]]s present in [[NLI]] and b) seem to be driven by the [[dominant discourse in the field]] at the [[time of their creation]]. </s>
+
In particular, [[we]] discuss the most common [[NLI resource]]s arguing that all these a) fail to capture the wealth of [[inferential mechanism]]s present in [[NLI]] and b) seem to be driven by the [[dominant discourse]] in the [[field]] at the [[time of their creation]]. </s>
In light of these [[observation]]s, we want to discuss the requirements that an adequate [[NLI platform]] must satisfy both in terms of the range of [[inference pattern]]s found in reasoning with [[NL]] as well as the range of the [[data collection mechanism]]s that are needed in order to acquire this range of [[inferential pattern]]s. </s>
+
In light of these [[observation]]s, we want to discuss the [[requirement]]s that an adequate [[NLI platform]] must satisfy both in terms of the range of [[inference pattern]]s found in reasoning with [[NL]] as well as the range of the [[data collection mechanism]]s that are needed in order to acquire this range of [[inferential pattern]]s. </s>
 +
=== 2 NLI Datasets ===
 +
In this section we present the [[merit]]s and [drawback]]s of three [[NLI dataset]]s which represent some of the main strategies that have been used in the [[data collection]] of [[inference]].
  
 
== References ==
 
== References ==

Revision as of 04:03, 13 September 2019

Subject Headings: Natural Language Inference Task.

Notes

Cited By

Quotes

Introduction

Understanding a Natural Language (NL) sentence amounts to understanding its consequences as well as the sentences that it is a consequence of. As Cooper et al. (1996) aptly put it "inferential ability is not only a central manifestation of semantic competence but is in fact centrally constitutive of it]]". Given the latter claim, it is no surprise that the same authors argue that the best way to test the semantic adequacy of NLP systems is to look at how well they perform with regard to Natural Language Inference (NLI). It is very hard not to agree with this statement. Indeed, NLI is considered by many researchers to be the crux of computational semantics. This paper is about the datasets created for this need. In particular, we discuss the most common NLI resources arguing that all these a) fail to capture the wealth of inferential mechanisms present in NLI and b) seem to be driven by the dominant discourse in the field at the time of their creation. In light of these observations, we want to discuss the requirements that an adequate NLI platform must satisfy both in terms of the range of inference patterns found in reasoning with NL as well as the range of the data collection mechanisms that are needed in order to acquire this range of inferential patterns.

2 NLI Datasets

In this section we present the merits and [drawback]]s of three NLI datasets which represent some of the main strategies that have been used in the data collection of inference.

References

;

 AuthorvolumeDate ValuetitletypejournaltitleUrldoinoteyear
2017 AnOverviewofNaturalLanguageInfeStergios Chatzikyriakidis
Robin Cooper
Simon Dobnik
Staffan Larsson
An Overview of Natural Language Inference Data Collection: The Way Forward?2017
AuthorStergios Chatzikyriakidis +, Robin Cooper +, Simon Dobnik + and Staffan Larsson +
titleAn Overview of Natural Language Inference Data Collection: The Way Forward? +
year2017 +