A Hybrid Deep Learning Approach to Detect Bangla Social Media Hate Speech

Tapotosh Ghosh; Ashraf Alam Khan Chowdhury; Md Hasan Al Banna; Md Jaber Al Nahian; M. Shamim Kaiser; Mufti Mahmud

doi:10.1007/978-981-19-2445-3_50

A Hybrid Deep Learning Approach to Detect Bangla Social Media Hate Speech

Tapotosh Ghosh^*
, Ashraf Alam Khan Chowdhury
, Md Hasan Al Banna
, Md Jaber Al Nahian
, M. Shamim Kaiser
, Mufti Mahmud

^*Corresponding author for this work

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

8 Scopus citations

Abstract

Social media has become an integral part of our day-to-day life. In our activities or posts on social media, the presence of hate speech written in the native language or English has increased significantly. It often leads to the spread of negativity, depression, or even sometimes considered cybercrime. In this paper, a hybrid deep learning approach has been taken to detect Bangla social media hate speech using fastText embedding, Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Convolutional Neural Network. A publicly available dataset of 30000 samples has been used, and the proposed hybrid model achieved close to 90% accuracy with significant sensitivity and specificity in Bangla hate speech detection. Several related deep learning approaches were evaluated in this same dataset, but none of them performed better than the proposed model. The hybrid model also showed robustness which made it more suitable for this task.

Original language	English
Title of host publication	Proceedings of International Conference on 4th Industrial Revolution and Beyond 2021
Editors	Sazzad Hossain, Md. Shahadat Hossain, M. Shamim Kaiser, Satya Prasad Majumder, Kanad Ray
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	711-722
Number of pages	12
ISBN (Print)	9789811924446
DOIs	https://doi.org/10.1007/978-981-19-2445-3_50
State	Published - 2022
Externally published	Yes
Event	1st International Conference on 4th Industrial Revolution and Beyond, IC4IR 2021 - Virtual, Online Duration: 10 Dec 2021 → 11 Dec 2021

Publication series

Name	Lecture Notes in Networks and Systems
Volume	437
ISSN (Print)	2367-3370
ISSN (Electronic)	2367-3389

Conference

Conference	1st International Conference on 4th Industrial Revolution and Beyond, IC4IR 2021
City	Virtual, Online
Period	10/12/21 → 11/12/21

Bibliographical note

Publisher Copyright:
© 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

Keywords

Embedding
Hate speech
Hybrid architecture
LSTM
Negativity

ASJC Scopus subject areas

Control and Systems Engineering
Signal Processing
Computer Networks and Communications

Access to Document

10.1007/978-981-19-2445-3_50

Cite this

Ghosh, T., Chowdhury, A. A. K., Banna, M. H. A., Nahian, M. J. A., Kaiser, M. S., & Mahmud, M. (2022). A Hybrid Deep Learning Approach to Detect Bangla Social Media Hate Speech. In S. Hossain, M. S. Hossain, M. S. Kaiser, S. P. Majumder, & K. Ray (Eds.), Proceedings of International Conference on 4th Industrial Revolution and Beyond 2021 (pp. 711-722). (Lecture Notes in Networks and Systems; Vol. 437). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-981-19-2445-3_50

Ghosh, Tapotosh ; Chowdhury, Ashraf Alam Khan ; Banna, Md Hasan Al et al. / A Hybrid Deep Learning Approach to Detect Bangla Social Media Hate Speech. Proceedings of International Conference on 4th Industrial Revolution and Beyond 2021. editor / Sazzad Hossain ; Md. Shahadat Hossain ; M. Shamim Kaiser ; Satya Prasad Majumder ; Kanad Ray. Springer Science and Business Media Deutschland GmbH, 2022. pp. 711-722 (Lecture Notes in Networks and Systems).

@inproceedings{fa691cb6cad34c02918335d2b252974b,

title = "A Hybrid Deep Learning Approach to Detect Bangla Social Media Hate Speech",

abstract = "Social media has become an integral part of our day-to-day life. In our activities or posts on social media, the presence of hate speech written in the native language or English has increased significantly. It often leads to the spread of negativity, depression, or even sometimes considered cybercrime. In this paper, a hybrid deep learning approach has been taken to detect Bangla social media hate speech using fastText embedding, Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Convolutional Neural Network. A publicly available dataset of 30000 samples has been used, and the proposed hybrid model achieved close to 90\% accuracy with significant sensitivity and specificity in Bangla hate speech detection. Several related deep learning approaches were evaluated in this same dataset, but none of them performed better than the proposed model. The hybrid model also showed robustness which made it more suitable for this task.",

keywords = "Embedding, Hate speech, Hybrid architecture, LSTM, Negativity",

author = "Tapotosh Ghosh and Chowdhury, \{Ashraf Alam Khan\} and Banna, \{Md Hasan Al\} and Nahian, \{Md Jaber Al\} and Kaiser, \{M. Shamim\} and Mufti Mahmud",

note = "Publisher Copyright: {\textcopyright} 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.; 1st International Conference on 4th Industrial Revolution and Beyond, IC4IR 2021 ; Conference date: 10-12-2021 Through 11-12-2021",

year = "2022",

doi = "10.1007/978-981-19-2445-3\_50",

language = "English",

isbn = "9789811924446",

series = "Lecture Notes in Networks and Systems",

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

pages = "711--722",

editor = "Sazzad Hossain and Hossain, \{Md. Shahadat\} and Kaiser, \{M. Shamim\} and Majumder, \{Satya Prasad\} and Kanad Ray",

booktitle = "Proceedings of International Conference on 4th Industrial Revolution and Beyond 2021",

address = "Germany",

}

Ghosh, T, Chowdhury, AAK, Banna, MHA, Nahian, MJA, Kaiser, MS & Mahmud, M 2022, A Hybrid Deep Learning Approach to Detect Bangla Social Media Hate Speech. in S Hossain, MS Hossain, MS Kaiser, SP Majumder & K Ray (eds), Proceedings of International Conference on 4th Industrial Revolution and Beyond 2021. Lecture Notes in Networks and Systems, vol. 437, Springer Science and Business Media Deutschland GmbH, pp. 711-722, 1st International Conference on 4th Industrial Revolution and Beyond, IC4IR 2021, Virtual, Online, 10/12/21. https://doi.org/10.1007/978-981-19-2445-3_50

A Hybrid Deep Learning Approach to Detect Bangla Social Media Hate Speech. / Ghosh, Tapotosh; Chowdhury, Ashraf Alam Khan; Banna, Md Hasan Al et al.
Proceedings of International Conference on 4th Industrial Revolution and Beyond 2021. ed. / Sazzad Hossain; Md. Shahadat Hossain; M. Shamim Kaiser; Satya Prasad Majumder; Kanad Ray. Springer Science and Business Media Deutschland GmbH, 2022. p. 711-722 (Lecture Notes in Networks and Systems; Vol. 437).

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

TY - GEN

T1 - A Hybrid Deep Learning Approach to Detect Bangla Social Media Hate Speech

AU - Ghosh, Tapotosh

AU - Chowdhury, Ashraf Alam Khan

AU - Banna, Md Hasan Al

AU - Nahian, Md Jaber Al

AU - Kaiser, M. Shamim

AU - Mahmud, Mufti

PY - 2022

Y1 - 2022

N2 - Social media has become an integral part of our day-to-day life. In our activities or posts on social media, the presence of hate speech written in the native language or English has increased significantly. It often leads to the spread of negativity, depression, or even sometimes considered cybercrime. In this paper, a hybrid deep learning approach has been taken to detect Bangla social media hate speech using fastText embedding, Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Convolutional Neural Network. A publicly available dataset of 30000 samples has been used, and the proposed hybrid model achieved close to 90% accuracy with significant sensitivity and specificity in Bangla hate speech detection. Several related deep learning approaches were evaluated in this same dataset, but none of them performed better than the proposed model. The hybrid model also showed robustness which made it more suitable for this task.

AB - Social media has become an integral part of our day-to-day life. In our activities or posts on social media, the presence of hate speech written in the native language or English has increased significantly. It often leads to the spread of negativity, depression, or even sometimes considered cybercrime. In this paper, a hybrid deep learning approach has been taken to detect Bangla social media hate speech using fastText embedding, Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Convolutional Neural Network. A publicly available dataset of 30000 samples has been used, and the proposed hybrid model achieved close to 90% accuracy with significant sensitivity and specificity in Bangla hate speech detection. Several related deep learning approaches were evaluated in this same dataset, but none of them performed better than the proposed model. The hybrid model also showed robustness which made it more suitable for this task.

KW - Embedding

KW - Hate speech

KW - Hybrid architecture

KW - LSTM

KW - Negativity

UR - https://www.scopus.com/pages/publications/85140740232

U2 - 10.1007/978-981-19-2445-3_50

DO - 10.1007/978-981-19-2445-3_50

M3 - Conference contribution

AN - SCOPUS:85140740232

SN - 9789811924446

T3 - Lecture Notes in Networks and Systems

SP - 711

EP - 722

BT - Proceedings of International Conference on 4th Industrial Revolution and Beyond 2021

A2 - Hossain, Sazzad

A2 - Hossain, Md. Shahadat

A2 - Kaiser, M. Shamim

A2 - Majumder, Satya Prasad

A2 - Ray, Kanad

PB - Springer Science and Business Media Deutschland GmbH

T2 - 1st International Conference on 4th Industrial Revolution and Beyond, IC4IR 2021

Y2 - 10 December 2021 through 11 December 2021

ER -

Ghosh T, Chowdhury AAK, Banna MHA, Nahian MJA, Kaiser MS, Mahmud M. A Hybrid Deep Learning Approach to Detect Bangla Social Media Hate Speech. In Hossain S, Hossain MS, Kaiser MS, Majumder SP, Ray K, editors, Proceedings of International Conference on 4th Industrial Revolution and Beyond 2021. Springer Science and Business Media Deutschland GmbH. 2022. p. 711-722. (Lecture Notes in Networks and Systems). doi: 10.1007/978-981-19-2445-3_50