BenCoDiC: A Large-Scale Contextual Cyberbullying Corpus for Bengali Political News Discourse FULL VERSION

Md. Nesarul Hoque; Rudra Pratap Deb Nath; Umme Salma; Shima Chakraborty; Md. Hanif Seddiqui

cs.CLOct 9, 2026

BenCoDiC: A Large-Scale Contextual Cyberbullying Corpus for Bengali Political News Discourse FULL VERSION

Authors: Md. Nesarul Hoque, Rudra Pratap Deb Nath, Umme Salma, Shima Chakraborty, Md. Hanif Seddiqui

Organizations: Big Data, Information and Knowledge Engineering Lab, Department of Computer Science and Engineering, University of Chittagong, Chattogram-4331, Bangladesh · Department of Computer Science and Engineering, Gopalganj Science and Technology University, Gopalganj-8105, Bangladesh · Department of Computer Science and Engineering, Bangladesh University, Dhaka-1207, Bangladesh

Abstract

Cyberbullying is a growing concern in online communities, yet existing Bengali corpora remain limited in contextual information, annotation transparency, scale, accessibility, and support for multidimensional analysis, thereby hindering robust detection and comprehensive study of harasser behaviors and networks. To address these gaps, we introduce the Bengali Contextual and Diverse Cyberbullying (BenCoDiC) corpus, a publicly available contextual cyberbullying resource for Bengali political news discourse comprising 50,250 textual instances collected from 100 YouTube videos and their associated discussions. The corpus preserves caption--comment--reply relationships and is organized through a nine-dimensional star schema covering users, videos, textual content, context, time, engagement, political parties, target identities, and language categories. Five AI chatbots were initially evaluated for annotation but did not meet the predefined reliability criterion. Consequently, a two-tier human annotation process with expert verification was adopted, achieving Fleiss' $κ=0.86$ and Krippendorff's $α=0.86$ . Statistical analyses characterize the corpus, while transformer-based experiments show that incorporating conversational context yields modest but consistent improvements over isolated-text classification. BenCoDiC supports context-aware cyberbullying detection and perpetrator-oriented behavioral and network analysis. The corpus is available at https://github.com/bencodic/BenCoDiC-Corpus.

Figures & tables

Reference

Task

Source

Cls

RG1

RG2

RG3

CHP

CnM

CAE

ARM

Size

Romim et al. [ 13 ]

Hate speech

Facebook, YouTube

✗

✓

30000

✓

Islam et al. [ 20 ]

Abusive

Facebook, YouTube

✗

12028

✗

Jahan et al. [ 22 ]

Abusive

Facebook

Bn, BnEn, RBn

✗

✓

✗

2000

✗

Chakraborty and Seddiqui [ 23 ]

Threat, abusive

Facebook

✗

5644

✗

Bhattacharya et al. [ 27 ]

Aggression, Misogyny

YouTube

Bn, Hn, En

BC, MC

✗

✓

>

25,000

✓

TABLE I : Comparison of Bengali corpora for cyberbullying and related social-media text classification tasks.

Fig. 1 : Development process of the BenCoDiC corpus.

Fig. 2 : The star schema of the BenCoDiC corpus.

Fact and	Attribute/	Description
Dimension	Column
	userId	Persistent anonymized identifier of the author’s YouTube channel
	handler	Anonymized handler of the author’s YouTube channel
	title	Anonymized title of the author’s YouTube channel
User	subscriberCount	Total number of channel subscribers
	videoCount	Total number of videos uploaded by the channel

TABLE II : Attributes and descriptions of the BenCoDiC star schema.

AI	Accuracy (%)				Mean	95% CI (%)
Chatbot	E-1	E-2	E-3	E-4	(%)	LB	UB
ChatGPT	71.00	71.40	69.80	76.60	72.20	67.41	76.99
Grok 3	80.2	75.8	82.0	81.0	79.75	75.40	84.10
DeepSeek	76.0	75.6	78.0	74.4	76.00	73.62	78.38
Gemini	71.4	70.4	69.6	71.2	70.65	69.34	71.96
Perplexity	70.6	69.6	68.0	66.0	68.55	65.35	71.75

TABLE III : Accuracy and 95% run-level CIs of free-tier AI chatbots for Bengali cyberbullying annotation.

Fig. 3 : Annotation process of the BenCoDiC corpus.

Measure	Level-1	Level-2
Fleiss’ $\kappa$	0.81	0.86
Krippendorff’s $\alpha$	0.81	0.86
Unanimous decisions (3–0), $n$ (%)	43,286 (86.14)	45,158 (89.87)
Majority decisions (2–1), $n$ (%)	6,964 (13.86)	5,092 (10.13)

TABLE IV : Inter-annotator agreement and decision statistics at Level-1 and Level-2.

Text	Class
(You were also a sycophant)	Bully
(I want proper justice)	Not Bully
(That bloody cheat!)	Bully
(Thanks to the student community)	Not Bully

TABLE V : Examples of the cyberbullying classes in BenCoDiC.

Class	min	max	avg	total	total
	Word	Word	Word	Word	Sample
Bully	1	620	12.52	376,655	30,083
Not Bully	1	629	9.41	189,780	20,167
Bully+Not Bully	1	629	11.27	566,435	50,250

TABLE VI : Textual statistics of the BenCoDiC corpus across cyberbullying classes.

Fig. 4 : Empirical cumulative distributions of text length across the Bully and Not Bully classes. The horizontal axis is presented on a logarithmic scale to accommodate the long-tailed word-count distributions.

Measure	Bully	Not Bully
Median words per sample	8	5
Vocabulary size	44,980	26,157
Shared vocabulary	13,709
Combined unique vocabulary	57,428
Jaccard vocabulary overlap	23.87%
Top five associated unigrams

TABLE VII : Additional text-length and class-associated lexical statistics.

Fig. 5 : Distribution of Bully and Not Bully samples across content types.

Engagement Measure	Mean	Median (IQR)	Mean	Median (IQR)	Spearman’s $\rho$	$\mathbf{p}$
	Bully		Not Bully
Likes	1.62	0 (0–0)	3.35	0 (0–1)	$-0.1314$	$<0.001$
Replies	0.10	0 (0–0)	0.18	0 (0–0)	$-0.0628$	$<0.001$

TABLE VIII : Engagement characteristics and their association with bullying labels for comments.

Month	Samples	Bully	Not Bully	Bully (%)
Jul. 2024	2,831	1,519	1,312	53.66
Aug. 2024	15,432	9,696	5,736	62.83
Sep. 2024	13,785	7,737	6,048	56.13
Oct. 2024	12,849	7,347	5,502	57.18
Nov. 2024	5,063	3,646	1,417	72.01
Dec. 2024	116	59	57	50.86

TABLE IX : Monthly distribution of labeled samples by posting month.

Fig. 6 : Monthly proportion of Bully samples by posting month.

News	Number	Number of Samples
Channel	of Videos	Bully	Not Bully	Total
Jamuna TV	12	4692	2967	7659
SOMOY TV	13	3373	2780	6153
Channel 24	13	3541	2612	6153
Ekattor TV	13	4208	1954	6162
Independent Television	13	4403	1956	6359

TABLE X : Channel-wise distribution of videos and labeled samples in the BenCoDiC corpus.

Political	Number	Number of Samples			Bully
Party	of Videos	Bully	Not Bully	Total	(%)
AL	16	5142	2826	7968	64.53
BNP	16	4573	3377	7950	57.52
JaPa	16	6404	1838	8242	77.70
Jamaat	16	3474	4432	7906	43.94
NCP	16	4977	3087	8064	61.72

TABLE XI : Distribution of videos and labeled samples across political-party categories.

Fig. 7 : Distribution of Bully and Not Bully samples across target identities.

Fig. 8 : Distribution of Bully and Not Bully samples across language categories.

	Bengali		Romanized Bengali		Code-Mix		English		Emoji-Only
Target Identity	Bully	Not Bully	Bully	Not Bully	Bully	Not Bully	Bully	Not Bully	Bully	Not Bully
Male	9,291	5,303	2,005	721	203	131	356	683	45	1,331
Female	1,865	655	343	98	64	23	72	88	3	117
Individual	172	139	56	45	5	4	5	26	0	36
People Group	4,018	2,000	579	232	101	56	80	86	0	40
Organization	8,815	4,449	1,122	403	280	205	229	445	10	302

TABLE XII : Cross-tabulation of target identity and language category across the Bully and Not Bully classes.

Split Seed	Partition	Samples	Videos	Captions	Comments	Replies	Bully	Not Bully	Bully (%)
	Train	30,104	60	60	26,568	3,476	18,024	12,080	59.87
0	Validation	10,180	20	20	8,888	1,272	6,133	4,047	60.25
	Test	9,966	20	20	8,912	1,034	5,926	4,040	59.46
	Train	30,201	60	60	26,617	3,524	17,879	12,322	59.20
42	Validation	9,952	20	20	8,775	1,157	6,042	3,910	60.71
	Test	10,097	20	20	8,976	1,101	6,162	3,935	61.03

TABLE XIII : Distribution of the video-disjoint training, validation, and test partitions across three split seeds.

Model	Setting	Macro Precision	Macro Recall	Macro F1	Weighted F1	Accuracy
BanglaBERT	Context-free	$0.8259\pm 0.0128$	$0.8315\pm 0.0086$	$0.8277\pm 0.0115$	$0.8343\pm 0.0115$	$0.8335\pm 0.0121$
	Flat context	$\mathbf{0.8359\pm 0.0092}$	$\mathbf{0.8351\pm 0.0087}$	$\mathbf{0.8353\pm 0.0085}$	$\mathbf{0.8425\pm 0.0078}$	$\mathbf{0.8426\pm 0.0080}$
Bangla-Bert-Base	Context-free	$0.7691\pm 0.0158$	$0.7681\pm 0.0141$	$0.7684\pm 0.0148$	$0.7786\pm 0.0138$	$0.7788\pm 0.0142$
	Flat context	$0.7758\pm 0.0135$	$0.7717\pm 0.0172$	$0.7732\pm 0.0156$	$0.7838\pm 0.0134$	$0.7847\pm 0.0125$
BanglishBERT	Context-free	$0.8163\pm 0.0138$	$0.8219\pm 0.0132$	$0.8184\pm 0.0138$	$0.8253\pm 0.0132$	$0.8245\pm 0.0134$
	Flat context	$0.8249\pm 0.0107$	$0.8258\pm 0.0118$	$0.8251\pm 0.0109$	$0.8325\pm 0.0102$	$0.8324\pm 0.0101$

TABLE XIV : Performance comparison of the context-free and flat-context settings. Results are reported as mean

\pm

standard deviation across nine evaluations.

Model	Mean $\Delta$ F1	95% CI	Significant
BanglaBERT	0.0076	$[0.0042,\ 0.0109]$	Yes
Bangla-Bert-Base	0.0048	$[0.0027,\ 0.0069]$	Yes
BanglishBERT	0.0067	$[0.0040,\ 0.0094]$	Yes
IndicBERT v2	0.0070	$[0.0038,\ 0.0103]$	Yes
mBERT	0.0084	$[0.0053,0.0114]$	Yes
XLM-R-base	0.0061	$[0.0038,\ 0.0085]$	Yes

TABLE XV : Paired analysis of the contextual gain in macro F1-score.

Configuration	Caption	Parent Comment	Macro F1	Weighted F1	Accuracy
A0	No	No	$0.7789\pm 0.0260$	$0.7821\pm 0.0280$	$0.7813\pm 0.0275$
A1	Yes	No	$0.7817\pm 0.0204$	$0.7848\pm 0.0216$	$0.7837\pm 0.0214$
A2	No	Yes	$0.7855\pm 0.0223$	$0.7888\pm 0.0234$	$0.7878\pm 0.0233$
A3	Yes	Yes	$\mathbf{0.7882\pm 0.0186}$	$\mathbf{0.7917\pm 0.0198}$	$\mathbf{0.7906\pm 0.0197}$

TABLE XVI : Reply-level context-element ablation results for BanglaBERT. Results are reported as mean

\pm

standard deviation across nine evaluations.

Bengali NLP

See all 23 papers

BanClickThumb: A Multimodal Dataset and Transformer Fusion Benchmarks for Clickbait Detection in Bengali YouTube Videos
Jul 19, 2026Md. Ariful Islam, Md Tanvirul Islam, Md. Maruf Hossain Miru +1Bengali NLP Multimodal Classification
Department of Computer Science and Engineering, Bangladesh Army University of Science and Technology, Saidpur Cantonment, Saidpur, Bangladesh
A Fusion of context-aware based BanglaBERT and Two-Layer Stacked LSTM Framework for Multi-Label Cyberbullying Detection
Feb 25, 2026Mirza Raquib, Asif Pervez Polok, Kedar Nath Biswas +3Bengali NLP Hybrid Transformer Architectures
Department of Computer and Communication Engineering, International Islamic University Chittagong, Chattogram, Bangladesh · Department of Information and Communication Engineering, Noakhali Science and Technology University, Noakhali, Bangladesh · mPower Social Enterprise +2
Ben-Misandry: A Bengali Dataset and Late-Fusion Transformer Framework for Detecting Anti-Male Gender Hate
Oct 17, 2026Md Mehedi Hasan Nipu, Chinmoy Mitra, Tarannum Ahmed Nowshin +1Hate Speech Detection Bengali NLP
Department of Electrical and Computer Engineering, North South University, Dhaka, Bangladesh · Department of Computer Science & Engineering, Rajshahi University of Engineering & Technology, Rajshahi, Bangladesh · Department of Computer Science and Engineering, BRAC University, Dhaka, Bangladesh +1

Jul 19, 2026J/K move · Enter open · S save

BanClickThumb: A Multimodal Dataset and Transformer Fusion Benchmarks for Clickbait Detection in Bengali YouTube Videos

Md. Ariful Islam, Md Tanvirul Islam, Md. Maruf Hossain Miru, Md Khalid Syfullah

Department of Computer Science and Engineering, Bangladesh Army University of Science and Technology, Saidpur Cantonment, Saidpur, Bangladesh

Bengali NLP Multimodal Classification Low-Resource Language Processing Multimodal Fusion

Open paper AskPDF arXiv

Clickbait, where video titles and thumbnails exaggerate or misrepresent content, reduces user trust, wastes attention, and promotes misinformation on video-sharing platforms. Detecting Bengali clickbait remains challenging because publicly available multimodal datasets are limited. To address this gap, we introduce BanClickThumb, a curated dataset of 7,147 Bengali YouTube thumbnail-title pairs from five content domains, annotated by ten annotators with high agreement (Cohen's Kappa: 0.83-0.93). Using this dataset, we benchmark text-only, image-only, and multimodal approaches. Among unimodal models, BanClickTextFormer (XLM-RoBERTa) achieves 0.82 accuracy, while BanClickImageFormer (SwiftFormer) reaches 0.68. Our proposed multimodal model, BanClickFusionFormer, combines ViT and XLM-RoBERTa through intermediate fusion and achieves the best accuracy of 0.84. Error analysis shows that dense thumbnail text, figurative language, and culturally specific slang remain challenging. Our findings demonstrate the effectiveness of multimodal fusion for Bengali clickbait detection and provide a publicly available benchmark to support future research on low-resource multimodal content analysis.

Abstract

Bengali NLP

Explore similar work

BanClickThumb: A Multimodal Dataset and Transformer Fusion Benchmarks for Clickbait Detection in Bengali YouTube Videos

A Fusion of context-aware based BanglaBERT and Two-Layer Stacked LSTM Framework for Multi-Label Cyberbullying Detection

Ben-Misandry: A Bengali Dataset and Late-Fusion Transformer Framework for Detecting Anti-Male Gender Hate