The Role of Output Vocabulary in T2T LMs for SPARQL Semantic Parsing

Debayan Banerjee; Pranav Nair; Ricardo Usbeck; Chris Biemann

doi:10.18653/v1/2023.findings-acl.774

The Role of Output Vocabulary in T2T LMs for SPARQL Semantic Parsing

Debayan Banerjee
, Pranav Nair
, Ricardo Usbeck
, Chris Biemann

Publikation: Beiträge in Sammelwerken › Aufsätze in Konferenzbänden › Forschung › Begutachtung

1 Zitat (Scopus)

Abstract

In this work, we analyse the role of output vocabulary for text-to-text (T2T) models on the task of SPARQL semantic parsing. We perform experiments within the the context of knowledge graph question answering (KGQA), where the task is to convert questions in natural language to the SPARQL query language. We observe that the query vocabulary is distinct from human vocabulary. Language Models (LMs) are pre-dominantly trained for human language tasks, and hence, if the query vocabulary is replaced with a vocabulary more attuned to the LM tokenizer, the performance of models may improve. We carry out carefully selected vocabulary substitutions on the queries and find absolute gains in the range of 17% on the GrailQA dataset.

Originalsprache	Englisch
Titel	Findings of the Association for Computational Linguistics: ACL 2023 : July 9-14, 2023
Redakteure/-innen	Anna Rogers, Jordan L. Boyd-Graber, Naoaki Okazaki
Seitenumfang	10
Erscheinungsort	Stroudsburg
Herausgeber (Verlag)	Association for Computational Linguistics (ACL)
Erscheinungsdatum	01.07.2023
Seiten	12219-12228
ISBN (elektronisch)	978-1-959429-62-3
DOIs	https://doi.org/10.18653/v1/2023.findings-acl.774 https://doi.org/10.48550/arXiv.2305.15108
Publikationsstatus	Erschienen - 01.07.2023
Extern publiziert	Ja
Veranstaltung	The 61st Annual Meeting of the Association for Computational Linguistics - Toronto, Kanada Dauer: 09.07.2023 → 14.07.2023 Konferenznummer: 61 https://2023.aclweb.org https://2023.aclweb.org/ https://dblp.org/streams/conf/acl#2023 http://www.wikidata.org/entity/Q119855443

Bibliographische Notiz

Publisher Copyright:
© 2023 Association for Computational Linguistics.

Fachgebiete und Schlagwörter

Wirtschaftsinformatik
Informatik

Zugriff auf Dokument

10.18653/v1/2023.findings-acl.774Lizenz: CC BY
10.48550/arXiv.2305.15108Lizenz: CC BY

Dieses zitieren

Banerjee, D., Nair, P., Usbeck, R., & Biemann, C. (2023). The Role of Output Vocabulary in T2T LMs for SPARQL Semantic Parsing. in A. Rogers, J. L. Boyd-Graber, & N. Okazaki (Hrsg.), Findings of the Association for Computational Linguistics: ACL 2023 : July 9-14, 2023 (S. 12219-12228). (Proceedings of the Annual Meeting of the Association for Computational Linguistics). Association for Computational Linguistics (ACL). https://doi.org/10.18653/v1/2023.findings-acl.774, https://doi.org/10.48550/arXiv.2305.15108

Banerjee, Debayan ; Nair, Pranav ; Usbeck, Ricardo et al. / The Role of Output Vocabulary in T2T LMs for SPARQL Semantic Parsing. Findings of the Association for Computational Linguistics: ACL 2023 : July 9-14, 2023. Hrsg. / Anna Rogers ; Jordan L. Boyd-Graber ; Naoaki Okazaki. Stroudsburg : Association for Computational Linguistics (ACL), 2023. S. 12219-12228 (Proceedings of the Annual Meeting of the Association for Computational Linguistics).

@inbook{5d6369c3cb2e4ab19990cad2ed7aa3f7,

title = "The Role of Output Vocabulary in T2T LMs for SPARQL Semantic Parsing",

abstract = "In this work, we analyse the role of output vocabulary for text-to-text (T2T) models on the task of SPARQL semantic parsing. We perform experiments within the the context of knowledge graph question answering (KGQA), where the task is to convert questions in natural language to the SPARQL query language. We observe that the query vocabulary is distinct from human vocabulary. Language Models (LMs) are pre-dominantly trained for human language tasks, and hence, if the query vocabulary is replaced with a vocabulary more attuned to the LM tokenizer, the performance of models may improve. We carry out carefully selected vocabulary substitutions on the queries and find absolute gains in the range of 17\% on the GrailQA dataset.",

keywords = "Business informatics, Informatics",

author = "Debayan Banerjee and Pranav Nair and Ricardo Usbeck and Chris Biemann",

note = "Publisher Copyright: {\textcopyright} 2023 Association for Computational Linguistics.; 61st Annual Meeting of the Association for Computational Linguistics, ACL 2023 ; Conference date: 09-07-2023 Through 14-07-2023",

year = "2023",

month = jul,

day = "1",

doi = "10.18653/v1/2023.findings-acl.774",

language = "English",

series = "Proceedings of the Annual Meeting of the Association for Computational Linguistics",

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

pages = "12219--12228",

editor = "Anna Rogers and Boyd-Graber, \{Jordan L.\} and Naoaki Okazaki",

booktitle = "Findings of the Association for Computational Linguistics: ACL 2023",

address = "United States",

url = "https://2023.aclweb.org, https://2023.aclweb.org/, https://dblp.org/streams/conf/acl\#2023, http://www.wikidata.org/entity/Q119855443",

}

Banerjee, D, Nair, P, Usbeck, R & Biemann, C 2023, The Role of Output Vocabulary in T2T LMs for SPARQL Semantic Parsing. in A Rogers, JL Boyd-Graber & N Okazaki (Hrsg.), Findings of the Association for Computational Linguistics: ACL 2023 : July 9-14, 2023. Proceedings of the Annual Meeting of the Association for Computational Linguistics, Association for Computational Linguistics (ACL), Stroudsburg, S. 12219-12228, The 61st Annual Meeting of the Association for Computational Linguistics, Toronto, Ontario, Kanada, 09.07.23. https://doi.org/10.18653/v1/2023.findings-acl.774, https://doi.org/10.48550/arXiv.2305.15108

The Role of Output Vocabulary in T2T LMs for SPARQL Semantic Parsing. / Banerjee, Debayan; Nair, Pranav; Usbeck, Ricardo et al.
Findings of the Association for Computational Linguistics: ACL 2023 : July 9-14, 2023. Hrsg. / Anna Rogers; Jordan L. Boyd-Graber; Naoaki Okazaki. Stroudsburg: Association for Computational Linguistics (ACL), 2023. S. 12219-12228 (Proceedings of the Annual Meeting of the Association for Computational Linguistics).

Publikation: Beiträge in Sammelwerken › Aufsätze in Konferenzbänden › Forschung › Begutachtung

TY - CHAP

T1 - The Role of Output Vocabulary in T2T LMs for SPARQL Semantic Parsing

AU - Banerjee, Debayan

AU - Nair, Pranav

AU - Usbeck, Ricardo

AU - Biemann, Chris

N1 - Conference code: 61

PY - 2023/7/1

Y1 - 2023/7/1

N2 - In this work, we analyse the role of output vocabulary for text-to-text (T2T) models on the task of SPARQL semantic parsing. We perform experiments within the the context of knowledge graph question answering (KGQA), where the task is to convert questions in natural language to the SPARQL query language. We observe that the query vocabulary is distinct from human vocabulary. Language Models (LMs) are pre-dominantly trained for human language tasks, and hence, if the query vocabulary is replaced with a vocabulary more attuned to the LM tokenizer, the performance of models may improve. We carry out carefully selected vocabulary substitutions on the queries and find absolute gains in the range of 17% on the GrailQA dataset.

AB - In this work, we analyse the role of output vocabulary for text-to-text (T2T) models on the task of SPARQL semantic parsing. We perform experiments within the the context of knowledge graph question answering (KGQA), where the task is to convert questions in natural language to the SPARQL query language. We observe that the query vocabulary is distinct from human vocabulary. Language Models (LMs) are pre-dominantly trained for human language tasks, and hence, if the query vocabulary is replaced with a vocabulary more attuned to the LM tokenizer, the performance of models may improve. We carry out carefully selected vocabulary substitutions on the queries and find absolute gains in the range of 17% on the GrailQA dataset.

KW - Business informatics

KW - Informatics

UR - https://www.scopus.com/pages/publications/85175442200

UR - https://www.mendeley.com/catalogue/ef5423a8-c326-3827-a9f1-bbd3d0e21ba1/

U2 - 10.18653/v1/2023.findings-acl.774

DO - 10.18653/v1/2023.findings-acl.774

M3 - Article in conference proceedings

T3 - Proceedings of the Annual Meeting of the Association for Computational Linguistics

SP - 12219

EP - 12228

BT - Findings of the Association for Computational Linguistics: ACL 2023

A2 - Rogers, Anna

A2 - Boyd-Graber, Jordan L.

A2 - Okazaki, Naoaki

PB - Association for Computational Linguistics (ACL)

CY - Stroudsburg

T2 - 61st Annual Meeting of the Association for Computational Linguistics

Y2 - 9 July 2023 through 14 July 2023

ER -

Banerjee D, Nair P, Usbeck R, Biemann C. The Role of Output Vocabulary in T2T LMs for SPARQL Semantic Parsing. in Rogers A, Boyd-Graber JL, Okazaki N, Hrsg., Findings of the Association for Computational Linguistics: ACL 2023 : July 9-14, 2023. Stroudsburg: Association for Computational Linguistics (ACL). 2023. S. 12219-12228. (Proceedings of the Annual Meeting of the Association for Computational Linguistics). doi: 10.18653/v1/2023.findings-acl.774, 10.48550/arXiv.2305.15108

The Role of Output Vocabulary in T2T LMs for SPARQL Semantic Parsing

Abstract

Bibliographische Notiz

Fachgebiete und Schlagwörter

Zugriff auf Dokument

Fingerprint

Dieses zitieren