QED: A Framework and Dataset for Explanations in Question Answering
Matthew Lamm,Jennimaria Palomaki,Chris Alberti,Daniel Andor,Eunsol Choi,Livio Baldini Soares,Michael Collins
Bạn đang xem: Business unusual
AbstractA question answering system that in addition to providing an answer provides an explanation of the reasoning that leads to lớn that answer has potential advantages in terms of debuggability, extensibility, và trust. Khổng lồ this end, we propose QED, a linguistically informed, extensible framework for explanations in question answering. A QED explanation specifies the relationship between a question và answer according to lớn formal semantic notions such as referential equality, sentencehood, and entailment. We describe & publicly release an expert-annotated dataset of QED explanations built upon a subset of the Google Natural Questions dataset, & report baseline models on two tasks—post- hoc explanation generation given an answer, và joint question answering & explanation generation. In the joint setting, a promising result suggests that training on a relatively small amount of QED data can improve question answering. In addition lớn describing the formal, language-theoretic motivations for the QED approach, we describe a large user study showing that the presence of QED explanations significantly improves the ability of untrained raters lớn spot errors made by a strong neural QA baseline.
Anthology ID:2021.tacl-1.48Volume:Transactions of the Association for Computational Linguistics, Volume 9Month:Year:2021Address:Cambridge, MAVenue:TACLSIG:Publisher:MIT PressNote:Pages:790–806Language:URL:https://slovenija-expo2000.com/2021.tacl-1.48DOI:10.1162/tacl_a_00398Bibkey:lamm-etal-2021-qedCite (ACL):Matthew Lamm, Jennimaria Palomaki, Chris Alberti, Daniel Andor, Eunsol Choi, Livio Baldini Soares, & Michael Collins. 2021. QED: A Framework and Dataset for Explanations in Question Answering. Transactions of the Association for Computational Linguistics, 9:790–806.Cite (Informal):QED: A Framework & Dataset for Explanations in Question Answering (Lamm et al., TACL 2021)Copy Citation:BibTeXMarkdownMODS XMLEndnoteMore options…PDF:https://slovenija-expo2000.com/2021.tacl-1.48.pdf
articlelamm-etal-2021-qed, title = "QED: A Framework and Dataset for Explanations in Question Answering", tác giả = "Lamm, Matthew và Palomaki, Jennimaria & Alberti, Chris & Andor, Daniel và Choi, Eunsol and Soares, Livio Baldini và Collins, Michael", journal = "Transactions of the Association for Computational Linguistics", volume = "9", year = "2021", address = "Cambridge, MA", advertiser = "MIT Press", url = "https://slovenija-expo2000.com/2021.tacl-1.48", doi = "10.1162/tacl_a_00398", pages = "790--806", abstract = "A question answering system that in addition lớn providing an answer provides an explanation of the reasoning that leads lớn that answer has potential advantages in terms of debuggability, extensibility, and trust. Khổng lồ this end, we propose QED, a linguistically informed, extensible framework for explanations in question answering. A QED explanation specifies the relationship between a question and answer according lớn formal semantic notions such as referential equality, sentencehood, và entailment. We describe & publicly release an expert-annotated dataset of QED explanations built upon a subset of the Google Natural Questions dataset, và report baseline models on two tasks---post- hoc explanation generation given an answer, & joint question answering và explanation generation. In the joint setting, a promising result suggests that training on a relatively small amount of QED data can improve question answering. In addition to describing the formal, language-theoretic motivations for the QED approach, we describe a large user study showing that the presence of QED explanations significantly improves the ability of untrained raters to lớn spot errors made by a strong neural QA baseline.",
QED: A Framework and Dataset for Explanations in Question Answering Matthew Lamm author Jennimaria Palomaki author Chris Alberti tác giả Daniel Andor author Eunsol Choi tác giả Livio Baldini Soares author Michael Collins author 2021 text journal article Transactions of the Association for Computational Linguistics continuing MIT Press Cambridge, MA periodical academic journal A question answering system that in addition to lớn providing an answer provides an explanation of the reasoning that leads to that answer has potential advantages in terms of debuggability, extensibility, & trust. Khổng lồ this end, we propose QED, a linguistically informed, extensible framework for explanations in question answering. A QED explanation specifies the relationship between a question & answer according khổng lồ formal semantic notions such as referential equality, sentencehood, and entailment. We describe và publicly release an expert-annotated dataset of QED explanations built upon a subset of the Google Natural Questions dataset, and report baseline models on two tasks—post- hoc explanation generation given an answer, & joint question answering and explanation generation. In the joint setting, a promising result suggests that training on a relatively small amount of QED data can improve question answering. In addition to describing the formal, language-theoretic motivations for the QED approach, we describe a large user study showing that the presence of QED explanations significantly improves the ability of untrained raters lớn spot errors made by a strong neural QA baseline. Lamm-etal-2021-qed 10.1162/tacl_a_00398 https://slovenija-expo2000.com/2021.tacl-1.48 2021 9 790 806
%0 Journal Article%T QED: A Framework and Dataset for Explanations in Question Answering%A Lamm, Matthew%A Palomaki, Jennimaria%A Alberti, Chris%A Andor, Daniel%A Choi, Eunsol%A Soares, Livio Baldini%A Collins, Michael%J Transactions of the Association for Computational Linguistics%D 2021%V 9%I MIT Press%C Cambridge, MA%F lamm-etal-2021-qed%X A question answering system that in addition to providing an answer provides an explanation of the reasoning that leads lớn that answer has potential advantages in terms of debuggability, extensibility, và trust. Lớn this end, we propose QED, a linguistically informed, extensible framework for explanations in question answering. A QED explanation specifies the relationship between a question & answer according khổng lồ formal semantic notions such as referential equality, sentencehood, & entailment. We describe & publicly release an expert-annotated dataset of QED explanations built upon a subset of the Google Natural Questions dataset, and report baseline models on two tasks—post- hoc explanation generation given an answer, & joint question answering & explanation generation. In the joint setting, a promising result suggests that training on a relatively small amount of QED data can improve question answering. In addition to describing the formal, language-theoretic motivations for the QED approach, we describe a large user study showing that the presence of QED explanations significantly improves the ability of untrained raters lớn spot errors made by a strong neural QA baseline.%9 journal article%R 10.1162/tacl_a_00398%U https://slovenija-expo2000.com/2021.tacl-1.48%U https://doi.org/10.1162/tacl_a_00398%P 790-806
Xem thêm: Thêm 0 02 Mol Naoh Vào Dung Dịch Chứa 0 01 Mol Crcl2
Markdown (Informal)