Feature-Rich Classifiers for Recognizing Textual Entailment in Indonesian

Rani Aulia Hidayat; Isnaini Nurul Khasanah; Wava Carissa Putri; Rahmad Mahendra

doi:10.1016/j.procs.2021.05.094

Feature-Rich Classifiers for Recognizing Textual Entailment in Indonesian

Rani Aulia Hidayat, Isnaini Nurul Khasanah, Wava Carissa Putri, Rahmad Mahendra

Faculty of Computer Science

Research output: Contribution to journal › Conference article › peer-review

3 Citations (Scopus)

Abstract

Recognizing Textual Entailment (RTE) is a Natural Language Processing task to determine whether a sentence (text) semantically entails another sentence (hypothesis). In this paper, we extracted and learned 35 features from a pair of text and hypothesis in Indonesian. The ablation study was conducted to analyze features contribution to RTE model. The experiments shown that, using Support Vector Machine (SVM) and Logistic Regression, the token-based features contribute positively to improve the model performance. The best model in our experiment is SVM that scored F1-Score of 79.65%. Despite sacrificing 5-points accuracy to the state-of-the-art BERT model, SVM classifier is 31 hours more efficient in terms of training time.

Original language	English
Pages (from-to)	148-155
Number of pages	8
Journal	Procedia CIRP
Volume	189
DOIs	https://doi.org/10.1016/j.procs.2021.05.094
Publication status	Published - 2021
Event	5th International Conference on Artificial Intelligence in Computational Linguistics, ACLing 2021 - Virtual, Online, United Arab Emirates Duration: 4 Jun 2021 → 5 Jun 2021

Keywords

Ablation study
Feature
Indonesia, Text classification
Natural Language Inference
Recognizing Textual Entailment

Access to Document

10.1016/j.procs.2021.05.094

Cite this

@article{54c467e4312c41b99de6c2f5ea3357ef,

title = "Feature-Rich Classifiers for Recognizing Textual Entailment in Indonesian",

abstract = "Recognizing Textual Entailment (RTE) is a Natural Language Processing task to determine whether a sentence (text) semantically entails another sentence (hypothesis). In this paper, we extracted and learned 35 features from a pair of text and hypothesis in Indonesian. The ablation study was conducted to analyze features contribution to RTE model. The experiments shown that, using Support Vector Machine (SVM) and Logistic Regression, the token-based features contribute positively to improve the model performance. The best model in our experiment is SVM that scored F1-Score of 79.65%. Despite sacrificing 5-points accuracy to the state-of-the-art BERT model, SVM classifier is 31 hours more efficient in terms of training time.",

keywords = "Ablation study, Feature, Indonesia, Text classification, Natural Language Inference, Recognizing Textual Entailment",

author = "Hidayat, {Rani Aulia} and Khasanah, {Isnaini Nurul} and Putri, {Wava Carissa} and Rahmad Mahendra",

note = "Funding Information: This research was supported by the research grant from Universitas Indonesia. Publisher Copyright: {\textcopyright} 2021 Elsevier B.V.. All rights reserved.; 5th International Conference on Artificial Intelligence in Computational Linguistics, ACLing 2021 ; Conference date: 04-06-2021 Through 05-06-2021",

year = "2021",

doi = "10.1016/j.procs.2021.05.094",

language = "English",

volume = "189",

pages = "148--155",

journal = "Procedia CIRP",

issn = "2212-8271",

publisher = "Elsevier BV",

}

TY - JOUR

T1 - Feature-Rich Classifiers for Recognizing Textual Entailment in Indonesian

AU - Hidayat, Rani Aulia

AU - Khasanah, Isnaini Nurul

AU - Putri, Wava Carissa

AU - Mahendra, Rahmad

PY - 2021

Y1 - 2021

N2 - Recognizing Textual Entailment (RTE) is a Natural Language Processing task to determine whether a sentence (text) semantically entails another sentence (hypothesis). In this paper, we extracted and learned 35 features from a pair of text and hypothesis in Indonesian. The ablation study was conducted to analyze features contribution to RTE model. The experiments shown that, using Support Vector Machine (SVM) and Logistic Regression, the token-based features contribute positively to improve the model performance. The best model in our experiment is SVM that scored F1-Score of 79.65%. Despite sacrificing 5-points accuracy to the state-of-the-art BERT model, SVM classifier is 31 hours more efficient in terms of training time.

AB - Recognizing Textual Entailment (RTE) is a Natural Language Processing task to determine whether a sentence (text) semantically entails another sentence (hypothesis). In this paper, we extracted and learned 35 features from a pair of text and hypothesis in Indonesian. The ablation study was conducted to analyze features contribution to RTE model. The experiments shown that, using Support Vector Machine (SVM) and Logistic Regression, the token-based features contribute positively to improve the model performance. The best model in our experiment is SVM that scored F1-Score of 79.65%. Despite sacrificing 5-points accuracy to the state-of-the-art BERT model, SVM classifier is 31 hours more efficient in terms of training time.

KW - Ablation study

KW - Feature

KW - Indonesia, Text classification

KW - Natural Language Inference

KW - Recognizing Textual Entailment

UR - http://www.scopus.com/inward/record.url?scp=85112814756&partnerID=8YFLogxK

U2 - 10.1016/j.procs.2021.05.094

DO - 10.1016/j.procs.2021.05.094

M3 - Conference article

AN - SCOPUS:85112814756

SN - 2212-8271

VL - 189

SP - 148

EP - 155

JO - Procedia CIRP

JF - Procedia CIRP

T2 - 5th International Conference on Artificial Intelligence in Computational Linguistics, ACLing 2021

Y2 - 4 June 2021 through 5 June 2021

ER -

Feature-Rich Classifiers for Recognizing Textual Entailment in Indonesian

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this