Generating and modifying natural language explanations

Abdus Salam; Rolf Schwitter; Mehmet A. Orgun

Generating and modifying natural language explanations

Abdus Salam, Rolf Schwitter, Mehmet A. Orgun

School of Computing

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

44 Downloads (Pure)

Abstract

HESIP is a hybrid explanation system for image predictions that combines sub-symbolic and symbolic machine learning techniques to explain the predictions of image classification tasks. The sub-symbolic component makes a prediction for an image and the symbolic component learns probabilistic symbolic rules in order to explain that prediction. In HESIP, the explanations are generated in controlled natural language from the learned probabilistic rules using a bi-directional logic grammar. In this paper, we present an explanation modification method where a human-in-the-loop can modify an incorrect explanation generated by the HESIP system and afterwards, the modified explanation is used by HESIP to learn a better explanation.

Original language	English
Title of host publication	Proceedings of the 19th Workshop of the Australasian Language Technology Association, ALTA 2021
Editors	Afshin Rahimi, William Lane, Guido Zuccon
Place of Publication	Stroudsburg, PA
Publisher	Association for Computational Linguistics (ACL)
Pages	149–157
Number of pages	9
Publication status	Published - 2021
Event	19th Annual Workshop of the Australasian Language Technology Association (ALTA 2021) - Online, Australia Duration: 8 Dec 2021 → 10 Dec 2021

Conference

Conference	19th Annual Workshop of the Australasian Language Technology Association (ALTA 2021)
Abbreviated title	ALTA 2021
Country/Territory	Australia
Period	8/12/21 → 10/12/21

Bibliographical note

Version archived for private and non-commercial use with the permission of the author/s and according to publisher conditions. For further rights please contact the publisher.

Access to Document

Publisher version (open access)Final published version, 774 KBLicence: CC BY

https://aclanthology.org/2021.alta-1.0

Cite this

@inproceedings{9aae7eb1f87a470b9172bbb726622f88,

title = "Generating and modifying natural language explanations",

abstract = "HESIP is a hybrid explanation system for image predictions that combines sub-symbolic and symbolic machine learning techniques to explain the predictions of image classification tasks. The sub-symbolic component makes a prediction for an image and the symbolic component learns probabilistic symbolic rules in order to explain that prediction. In HESIP, the explanations are generated in controlled natural language from the learned probabilistic rules using a bi-directional logic grammar. In this paper, we present an explanation modification method where a human-in-the-loop can modify an incorrect explanation generated by the HESIP system and afterwards, the modified explanation is used by HESIP to learn a better explanation.",

author = "Abdus Salam and Rolf Schwitter and Orgun, {Mehmet A.}",

note = "Version archived for private and non-commercial use with the permission of the author/s and according to publisher conditions. For further rights please contact the publisher. ; 19th Annual Workshop of the Australasian Language Technology Association (ALTA 2021), ALTA 2021 ; Conference date: 08-12-2021 Through 10-12-2021",

year = "2021",

language = "English",

pages = "149–157",

editor = "Afshin Rahimi and William Lane and Guido Zuccon",

booktitle = "Proceedings of the 19th Workshop of the Australasian Language Technology Association, ALTA 2021",

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

}

Salam, A, Schwitter, R & Orgun, MA 2021, Generating and modifying natural language explanations. in A Rahimi, W Lane & G Zuccon (eds), Proceedings of the 19th Workshop of the Australasian Language Technology Association, ALTA 2021. Association for Computational Linguistics (ACL), Stroudsburg, PA, pp. 149–157, 19th Annual Workshop of the Australasian Language Technology Association (ALTA 2021), Queensland, Australia, 8/12/21. <https://aclanthology.org/2021.alta-1.0>

Generating and modifying natural language explanations. / Salam, Abdus; Schwitter, Rolf ; Orgun, Mehmet A.
Proceedings of the 19th Workshop of the Australasian Language Technology Association, ALTA 2021. ed. / Afshin Rahimi; William Lane; Guido Zuccon. Stroudsburg, PA: Association for Computational Linguistics (ACL), 2021. p. 149–157.

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

TY - GEN

T1 - Generating and modifying natural language explanations

AU - Salam, Abdus

AU - Schwitter, Rolf

AU - Orgun, Mehmet A.

N1 - Version archived for private and non-commercial use with the permission of the author/s and according to publisher conditions. For further rights please contact the publisher.

PY - 2021

Y1 - 2021

N2 - HESIP is a hybrid explanation system for image predictions that combines sub-symbolic and symbolic machine learning techniques to explain the predictions of image classification tasks. The sub-symbolic component makes a prediction for an image and the symbolic component learns probabilistic symbolic rules in order to explain that prediction. In HESIP, the explanations are generated in controlled natural language from the learned probabilistic rules using a bi-directional logic grammar. In this paper, we present an explanation modification method where a human-in-the-loop can modify an incorrect explanation generated by the HESIP system and afterwards, the modified explanation is used by HESIP to learn a better explanation.

AB - HESIP is a hybrid explanation system for image predictions that combines sub-symbolic and symbolic machine learning techniques to explain the predictions of image classification tasks. The sub-symbolic component makes a prediction for an image and the symbolic component learns probabilistic symbolic rules in order to explain that prediction. In HESIP, the explanations are generated in controlled natural language from the learned probabilistic rules using a bi-directional logic grammar. In this paper, we present an explanation modification method where a human-in-the-loop can modify an incorrect explanation generated by the HESIP system and afterwards, the modified explanation is used by HESIP to learn a better explanation.

UR - http://www.scopus.com/inward/record.url?scp=85127113159&partnerID=8YFLogxK

M3 - Conference proceeding contribution

SP - 149

EP - 157

BT - Proceedings of the 19th Workshop of the Australasian Language Technology Association, ALTA 2021

A2 - Rahimi, Afshin

A2 - Lane, William

A2 - Zuccon, Guido

PB - Association for Computational Linguistics (ACL)

CY - Stroudsburg, PA

T2 - 19th Annual Workshop of the Australasian Language Technology Association (ALTA 2021)

Y2 - 8 December 2021 through 10 December 2021

ER -

Generating and modifying natural language explanations

Abstract

Conference

Bibliographical note

Access to Document

Other files and links

Fingerprint

Cite this