Tensor Space Model-based Textual Data Augmentation for Text Classification

Minsuk Chang; Han Joon Kim

doi:10.1109/BigData59044.2023.10386629

Tensor Space Model-based Textual Data Augmentation for Text Classification

Minsuk Chang, Han Joon Kim

School of Electrical and Computer Engineering

Seoul National University

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

Abstract

In this paper, we first introduce a new text representation method to convert a textual document into a tensor space model named TextCuboid, which can preserve various meanings of polysemy. Based upon the new model, we propose two novel data augmentation techniques (called Boolean augmentation and CuboidGAN) that can be directly applied to the TextCuboid model for text classification tasks. Boolean augmentation includes three simple keyword modifications: synonym replacement, synonym insertion, and random deletion. CuboidGAN is composed of two key components, style encoding, and residual regression, and it is trained in two phases to generate unambiguous and plausible concept vectors. Through intensive experiments using five commonly used datasets, we prove that our proposed methods perform better data augmentation than other conventional methods. We also show that each augmentation method component significantly contributes to text classification through ablation studies.

Original language	English
Title of host publication	Proceedings - 2023 IEEE International Conference on Big Data, BigData 2023
Editors	Jingrui He, Themis Palpanas, Xiaohua Hu, Alfredo Cuzzocrea, Dejing Dou, Dominik Slezak, Wei Wang, Aleksandra Gruca, Jerry Chun-Wei Lin, Rakesh Agrawal
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	4276-4283
Number of pages	8
ISBN (Electronic)	9798350324457
DOIs	https://doi.org/10.1109/BigData59044.2023.10386629
State	Published - 2023
Event	2023 IEEE International Conference on Big Data, BigData 2023 - Sorrento, Italy Duration: 15 Dec 2023 → 18 Dec 2023

Publication series

Name	Proceedings - 2023 IEEE International Conference on Big Data, BigData 2023

Conference

Conference	2023 IEEE International Conference on Big Data, BigData 2023
Country/Territory	Italy
City	Sorrento
Period	15/12/23 → 18/12/23

Keywords

Autoencoder
Data Augmentation
Deep Learning
Generative Adversarial Networks
Tensor Space Model
Text Classification
Text Representation Model

Access to Document

10.1109/BigData59044.2023.10386629

Cite this

Chang, M., & Kim, H. J. (2023). Tensor Space Model-based Textual Data Augmentation for Text Classification. In J. He, T. Palpanas, X. Hu, A. Cuzzocrea, D. Dou, D. Slezak, W. Wang, A. Gruca, J. C.-W. Lin, & R. Agrawal (Eds.), Proceedings - 2023 IEEE International Conference on Big Data, BigData 2023 (pp. 4276-4283). (Proceedings - 2023 IEEE International Conference on Big Data, BigData 2023). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/BigData59044.2023.10386629

Chang, Minsuk ; Kim, Han Joon. / Tensor Space Model-based Textual Data Augmentation for Text Classification. Proceedings - 2023 IEEE International Conference on Big Data, BigData 2023. editor / Jingrui He ; Themis Palpanas ; Xiaohua Hu ; Alfredo Cuzzocrea ; Dejing Dou ; Dominik Slezak ; Wei Wang ; Aleksandra Gruca ; Jerry Chun-Wei Lin ; Rakesh Agrawal. Institute of Electrical and Electronics Engineers Inc., 2023. pp. 4276-4283 (Proceedings - 2023 IEEE International Conference on Big Data, BigData 2023).

@inproceedings{245cee0849ca4298b4715f2e729f979b,

title = "Tensor Space Model-based Textual Data Augmentation for Text Classification",

abstract = "In this paper, we first introduce a new text representation method to convert a textual document into a tensor space model named TextCuboid, which can preserve various meanings of polysemy. Based upon the new model, we propose two novel data augmentation techniques (called Boolean augmentation and CuboidGAN) that can be directly applied to the TextCuboid model for text classification tasks. Boolean augmentation includes three simple keyword modifications: synonym replacement, synonym insertion, and random deletion. CuboidGAN is composed of two key components, style encoding, and residual regression, and it is trained in two phases to generate unambiguous and plausible concept vectors. Through intensive experiments using five commonly used datasets, we prove that our proposed methods perform better data augmentation than other conventional methods. We also show that each augmentation method component significantly contributes to text classification through ablation studies.",

keywords = "Autoencoder, Data Augmentation, Deep Learning, Generative Adversarial Networks, Tensor Space Model, Text Classification, Text Representation Model",

author = "Minsuk Chang and Kim, {Han Joon}",

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; 2023 IEEE International Conference on Big Data, BigData 2023 ; Conference date: 15-12-2023 Through 18-12-2023",

year = "2023",

doi = "10.1109/BigData59044.2023.10386629",

language = "English",

series = "Proceedings - 2023 IEEE International Conference on Big Data, BigData 2023",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "4276--4283",

editor = "Jingrui He and Themis Palpanas and Xiaohua Hu and Alfredo Cuzzocrea and Dejing Dou and Dominik Slezak and Wei Wang and Aleksandra Gruca and Lin, {Jerry Chun-Wei} and Rakesh Agrawal",

booktitle = "Proceedings - 2023 IEEE International Conference on Big Data, BigData 2023",

address = "United States",

}

Chang, M & Kim, HJ 2023, Tensor Space Model-based Textual Data Augmentation for Text Classification. in J He, T Palpanas, X Hu, A Cuzzocrea, D Dou, D Slezak, W Wang, A Gruca, JC-W Lin & R Agrawal (eds), Proceedings - 2023 IEEE International Conference on Big Data, BigData 2023. Proceedings - 2023 IEEE International Conference on Big Data, BigData 2023, Institute of Electrical and Electronics Engineers Inc., pp. 4276-4283, 2023 IEEE International Conference on Big Data, BigData 2023, Sorrento, Italy, 15/12/23. https://doi.org/10.1109/BigData59044.2023.10386629

Tensor Space Model-based Textual Data Augmentation for Text Classification. / Chang, Minsuk; Kim, Han Joon.
Proceedings - 2023 IEEE International Conference on Big Data, BigData 2023. ed. / Jingrui He; Themis Palpanas; Xiaohua Hu; Alfredo Cuzzocrea; Dejing Dou; Dominik Slezak; Wei Wang; Aleksandra Gruca; Jerry Chun-Wei Lin; Rakesh Agrawal. Institute of Electrical and Electronics Engineers Inc., 2023. p. 4276-4283 (Proceedings - 2023 IEEE International Conference on Big Data, BigData 2023).

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

TY - GEN

T1 - Tensor Space Model-based Textual Data Augmentation for Text Classification

AU - Chang, Minsuk

AU - Kim, Han Joon

PY - 2023

Y1 - 2023

N2 - In this paper, we first introduce a new text representation method to convert a textual document into a tensor space model named TextCuboid, which can preserve various meanings of polysemy. Based upon the new model, we propose two novel data augmentation techniques (called Boolean augmentation and CuboidGAN) that can be directly applied to the TextCuboid model for text classification tasks. Boolean augmentation includes three simple keyword modifications: synonym replacement, synonym insertion, and random deletion. CuboidGAN is composed of two key components, style encoding, and residual regression, and it is trained in two phases to generate unambiguous and plausible concept vectors. Through intensive experiments using five commonly used datasets, we prove that our proposed methods perform better data augmentation than other conventional methods. We also show that each augmentation method component significantly contributes to text classification through ablation studies.

AB - In this paper, we first introduce a new text representation method to convert a textual document into a tensor space model named TextCuboid, which can preserve various meanings of polysemy. Based upon the new model, we propose two novel data augmentation techniques (called Boolean augmentation and CuboidGAN) that can be directly applied to the TextCuboid model for text classification tasks. Boolean augmentation includes three simple keyword modifications: synonym replacement, synonym insertion, and random deletion. CuboidGAN is composed of two key components, style encoding, and residual regression, and it is trained in two phases to generate unambiguous and plausible concept vectors. Through intensive experiments using five commonly used datasets, we prove that our proposed methods perform better data augmentation than other conventional methods. We also show that each augmentation method component significantly contributes to text classification through ablation studies.

KW - Autoencoder

KW - Data Augmentation

KW - Deep Learning

KW - Generative Adversarial Networks

KW - Tensor Space Model

KW - Text Classification

KW - Text Representation Model

UR - http://www.scopus.com/inward/record.url?scp=85184987960&partnerID=8YFLogxK

U2 - 10.1109/BigData59044.2023.10386629

DO - 10.1109/BigData59044.2023.10386629

M3 - Conference contribution

AN - SCOPUS:85184987960

T3 - Proceedings - 2023 IEEE International Conference on Big Data, BigData 2023

SP - 4276

EP - 4283

BT - Proceedings - 2023 IEEE International Conference on Big Data, BigData 2023

A2 - He, Jingrui

A2 - Palpanas, Themis

A2 - Hu, Xiaohua

A2 - Cuzzocrea, Alfredo

A2 - Dou, Dejing

A2 - Slezak, Dominik

A2 - Wang, Wei

A2 - Gruca, Aleksandra

A2 - Lin, Jerry Chun-Wei

A2 - Agrawal, Rakesh

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2023 IEEE International Conference on Big Data, BigData 2023

Y2 - 15 December 2023 through 18 December 2023

ER -

Chang M, Kim HJ. Tensor Space Model-based Textual Data Augmentation for Text Classification. In He J, Palpanas T, Hu X, Cuzzocrea A, Dou D, Slezak D, Wang W, Gruca A, Lin JCW, Agrawal R, editors, Proceedings - 2023 IEEE International Conference on Big Data, BigData 2023. Institute of Electrical and Electronics Engineers Inc. 2023. p. 4276-4283. (Proceedings - 2023 IEEE International Conference on Big Data, BigData 2023). doi: 10.1109/BigData59044.2023.10386629

Tensor Space Model-based Textual Data Augmentation for Text Classification

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this