Semi-supervised Textual Entailment on Indonesian Wikipedia Data

Ken Nabila Setya; Rahmad Mahendra

doi:10.1007/978-3-031-23793-5_34

Semi-supervised Textual Entailment on Indonesian Wikipedia Data

Ken Nabila Setya, Rahmad Mahendra

Department of Computer Science

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

5 Citations (Scopus)

Abstract

Recognizing Textual Entailment (RTE) is a research in Natural Language Processing that aims to identify whether there is an entailment relation between two texts. Textual Entailment has been studied in a variety of languages, but it is rare for the Indonesian language. The purpose of the work presented in this paper is to conduct the RTE experiment on Indonesian language dataset. Since manual data creation is costly and time consuming, we choose semi-supervised machine learning approach. We apply co-training algorithm to enlarge small amounts of annotated data, called seeds. With this method, the human effort only needed to annotate the seeds. The data resource used is all from Wikipedia and the entailment pairs are extracted from its revision history. Evaluation of 1,857 sentence pairs labelled with entailment information using our approach achieve accuracy 76%.

Original language	English
Title of host publication	Computational Linguistics and Intelligent Text Processing - 19th International Conference, CICLing 2018, Revised Selected Papers
Editors	Alexander Gelbukh
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	416-427
Number of pages	12
ISBN (Print)	9783031237928
DOIs	https://doi.org/10.1007/978-3-031-23793-5_34
Publication status	Published - 2023
Event	19th International Conference on Computational Linguistics and Intelligent Text Processing, CICLing 2018 - Hanoi, Viet Nam Duration: 18 Mar 2018 → 24 Mar 2018

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	13396 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	19th International Conference on Computational Linguistics and Intelligent Text Processing, CICLing 2018
Country/Territory	Viet Nam
City	Hanoi
Period	18/03/18 → 24/03/18

Keywords

Co-training
Indonesian language
LSTM
Semi-supervised
Textual entailment
Wikipedia revision history

Access to Document

10.1007/978-3-031-23793-5_34

Cite this

Setya, K. N., & Mahendra, R. (2023). Semi-supervised Textual Entailment on Indonesian Wikipedia Data. In A. Gelbukh (Ed.), Computational Linguistics and Intelligent Text Processing - 19th International Conference, CICLing 2018, Revised Selected Papers (pp. 416-427). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13396 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-23793-5_34

Setya, Ken Nabila ; Mahendra, Rahmad. / Semi-supervised Textual Entailment on Indonesian Wikipedia Data. Computational Linguistics and Intelligent Text Processing - 19th International Conference, CICLing 2018, Revised Selected Papers. editor / Alexander Gelbukh. Springer Science and Business Media Deutschland GmbH, 2023. pp. 416-427 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{1020bee17eaa44f28a125cd876f1d375,

title = "Semi-supervised Textual Entailment on Indonesian Wikipedia Data",

abstract = "Recognizing Textual Entailment (RTE) is a research in Natural Language Processing that aims to identify whether there is an entailment relation between two texts. Textual Entailment has been studied in a variety of languages, but it is rare for the Indonesian language. The purpose of the work presented in this paper is to conduct the RTE experiment on Indonesian language dataset. Since manual data creation is costly and time consuming, we choose semi-supervised machine learning approach. We apply co-training algorithm to enlarge small amounts of annotated data, called seeds. With this method, the human effort only needed to annotate the seeds. The data resource used is all from Wikipedia and the entailment pairs are extracted from its revision history. Evaluation of 1,857 sentence pairs labelled with entailment information using our approach achieve accuracy 76%.",

keywords = "Co-training, Indonesian language, LSTM, Semi-supervised, Textual entailment, Wikipedia revision history",

author = "Setya, {Ken Nabila} and Rahmad Mahendra",

note = "Funding Information: Acknowledgement. We thank to Mirna Adriani for input and comments. This research was partially supported by PITTA UI Grant Contract No. 410/UN2.R3.1/HKP.05.00/ 2017. The first author was also partially funded by Bukalapak. Publisher Copyright: {\textcopyright} 2023, Springer Nature Switzerland AG.; 19th International Conference on Computational Linguistics and Intelligent Text Processing, CICLing 2018 ; Conference date: 18-03-2018 Through 24-03-2018",

year = "2023",

doi = "10.1007/978-3-031-23793-5_34",

language = "English",

isbn = "9783031237928",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "416--427",

editor = "Alexander Gelbukh",

booktitle = "Computational Linguistics and Intelligent Text Processing - 19th International Conference, CICLing 2018, Revised Selected Papers",

address = "Germany",

}

Setya, KN & Mahendra, R 2023, Semi-supervised Textual Entailment on Indonesian Wikipedia Data. in A Gelbukh (ed.), Computational Linguistics and Intelligent Text Processing - 19th International Conference, CICLing 2018, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 13396 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 416-427, 19th International Conference on Computational Linguistics and Intelligent Text Processing, CICLing 2018, Hanoi, Viet Nam, 18/03/18. https://doi.org/10.1007/978-3-031-23793-5_34

Semi-supervised Textual Entailment on Indonesian Wikipedia Data. / Setya, Ken Nabila; Mahendra, Rahmad.
Computational Linguistics and Intelligent Text Processing - 19th International Conference, CICLing 2018, Revised Selected Papers. ed. / Alexander Gelbukh. Springer Science and Business Media Deutschland GmbH, 2023. p. 416-427 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13396 LNCS).

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

TY - GEN

T1 - Semi-supervised Textual Entailment on Indonesian Wikipedia Data

AU - Setya, Ken Nabila

AU - Mahendra, Rahmad

N1 - Funding Information: Acknowledgement. We thank to Mirna Adriani for input and comments. This research was partially supported by PITTA UI Grant Contract No. 410/UN2.R3.1/HKP.05.00/ 2017. The first author was also partially funded by Bukalapak. Publisher Copyright: © 2023, Springer Nature Switzerland AG.

PY - 2023

Y1 - 2023

N2 - Recognizing Textual Entailment (RTE) is a research in Natural Language Processing that aims to identify whether there is an entailment relation between two texts. Textual Entailment has been studied in a variety of languages, but it is rare for the Indonesian language. The purpose of the work presented in this paper is to conduct the RTE experiment on Indonesian language dataset. Since manual data creation is costly and time consuming, we choose semi-supervised machine learning approach. We apply co-training algorithm to enlarge small amounts of annotated data, called seeds. With this method, the human effort only needed to annotate the seeds. The data resource used is all from Wikipedia and the entailment pairs are extracted from its revision history. Evaluation of 1,857 sentence pairs labelled with entailment information using our approach achieve accuracy 76%.

AB - Recognizing Textual Entailment (RTE) is a research in Natural Language Processing that aims to identify whether there is an entailment relation between two texts. Textual Entailment has been studied in a variety of languages, but it is rare for the Indonesian language. The purpose of the work presented in this paper is to conduct the RTE experiment on Indonesian language dataset. Since manual data creation is costly and time consuming, we choose semi-supervised machine learning approach. We apply co-training algorithm to enlarge small amounts of annotated data, called seeds. With this method, the human effort only needed to annotate the seeds. The data resource used is all from Wikipedia and the entailment pairs are extracted from its revision history. Evaluation of 1,857 sentence pairs labelled with entailment information using our approach achieve accuracy 76%.

KW - Co-training

KW - Indonesian language

KW - LSTM

KW - Semi-supervised

KW - Textual entailment

KW - Wikipedia revision history

UR - http://www.scopus.com/inward/record.url?scp=85149624818&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-23793-5_34

DO - 10.1007/978-3-031-23793-5_34

M3 - Conference contribution

AN - SCOPUS:85149624818

SN - 9783031237928

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 416

EP - 427

BT - Computational Linguistics and Intelligent Text Processing - 19th International Conference, CICLing 2018, Revised Selected Papers

A2 - Gelbukh, Alexander

PB - Springer Science and Business Media Deutschland GmbH

T2 - 19th International Conference on Computational Linguistics and Intelligent Text Processing, CICLing 2018

Y2 - 18 March 2018 through 24 March 2018

ER -

Setya KN, Mahendra R. Semi-supervised Textual Entailment on Indonesian Wikipedia Data. In Gelbukh A, editor, Computational Linguistics and Intelligent Text Processing - 19th International Conference, CICLing 2018, Revised Selected Papers. Springer Science and Business Media Deutschland GmbH. 2023. p. 416-427. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-23793-5_34