Modeling temporal patterns of cyberbullying detection with hierarchical attention networks
Open accessAbstract
Cyberbullying is rapidly becoming one of the most serious online risks for adolescents. This has motivated work on machine learning methods to automate the process of cyberbullying detection, which have so far mostly viewed cyberbullying as one-off incidents that occur at a single point in time. Comparatively less is known about how cyberbullying behavior occurs and evolves over time. This oversight highlights a crucial open challenge for cyberbullying-related research, given that cyberbullying is typically defined as intentional acts of aggression via electronic communication that occur repeatedly and persistently . In this article, we center our discussion on the challenge of modeling temporal patterns of cyberbullying behavior. Specifically, we investigate how temporal information within a social media session, which has an inherently hierarchical structure (e.g., words form a comment and comments form a session), can be leveraged to facilitate cyberbullying detection. Recent findings from interdisciplinary research suggest that the temporal characteristics of bullying sessions differ from those of non-bullying sessions and that the temporal information from users’ comments can improve cyberbullying detection. The proposed framework consists of three distinctive features: (1) a hierarchical structure that reflects how a social media session is formed in a bottom-up manner; (2) attention mechanisms applied at the word- and comment-level to differentiate the contributions of words and comments to the representation of a social media session; and (3) the incorporation of temporal features in modeling cyberbullying behavior at the comment-level. Quantitative and qualitative evaluations are conducted on a real-world dataset collected from Instagram, the social networking site with the highest percentage of users reporting cyberbullying experiences. Results from empirical evaluations show the significance of the proposed methods, which are tailored to capture temporal patterns of cyberbullying detection.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
HANs have been used to understand the common tone of a social media message and the presence of certain bullying behaviors at the sentence level.
“HANs have been used to understand the common tone of a social media message and the presence of certain bullying behaviors at the sentence level [61,118,119].”
From A Review on Deep-Learning-Based Cyberbullying Detection (2023) · cites this paper as [119]
The Proposed Framework (§4) describes word- and comment-level attention, and the Case Study (§5.6) shows that the network highlights words associated with bullying. The paper classifies whole social media sessions, however; these passages do not establish that it determines a message’s common tone or identifies bullying behaviors at the sentence level. Page numbers were not provided in our copy. (LLM, conf=0.86)
Experimental — accuracy not yet measuredHierarchical Attention Networks (HAN) have demonstrated promising results at both the document and sentence levels, especially in detecting cyberbullying.
“Hierarchical Attention Networks (HAN) have demonstrated promising results at both the document and sentence levels, especially in detecting cyberbullying.”
From A Review on Deep-Learning-Based Cyberbullying Detection (2023) · cites this paper as [119]
Section 2.2 reports promising results for HAN in document classification and describes word- and sentence-level attention. Section 5.3.1 reports that HAN-based models performed well at identifying cyberbullying sessions. The paper does not report separate sentence-level detection results; its cyberbullying results are at the session level. (LLM, conf=0.85)
Experimental — accuracy not yet measuredMost cyberbullying detection algorithms are developed for high-resource languages like English, Chinese, French, and German, causing bias in detecting low-resource languages like Swahili, Bengali, and Marathi, which may affect accuracy, precision, and fairness.
“(2) Mostly of cyberbullying detection algorithms are developed to capture cyberbullying instances for high resource languages like English, Chinese, French, and Germany, so it causes bias in detecting cyberbullying instances on low resource languages like Swahili, Bengali, and Marathi which may affect the accuracy, precision and fairness of cyberbullying classification processes [7].”
From Cyberbullying detection: Exploring datasets, technologies, and approaches on social media platforms (2026) · cites this paper as [7]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
The citation lists haven’t been fetched yet
References 52 · Cited by 43
Research works printed but not linked: 8
- Lu Cheng, Jundong Li, Yasin N. Silva, Deborah L. Hall, and Huan Liu. 2019. XBully: Cyberbullying detection within a multi-modal context. In Proceedings of the 12th ACM International Conference on Web Search and Data Mining. 339- 347.
- Alexis Conneau, Holger Schwenk, Loïc Barrault, and Yann Lecun. 2016. Very deep convolutional networks for text classification. arXiv preprint arXiv:1606.01781 (2016). · arXiv
- Stefan Kombrink, Tomáš Mikolov, Martin Karafiát, and Lukáš Burget. 2011. Recurrent neural network based language modeling in meeting recognition. In Proceedings of the 12th Conference of the International Speech Communication Association.
- Vinita Nahar, Xue Li, and Chaoyi Pang. 2013. An effective approach for cyberbullying detection. Commun. Inf. Sci. Manag. Eng. 3, 5 (2013), 238.
- James W. Pennebaker, Martha E. Francis, and Roger J. Booth. 2001. Linguistic inquiry and word count: LIWC 2001. Mahway: Lawr. Erlb. Assoc. 71, 2001 (2001).
- Zhilin Yang, Zihang Dai, Yiming Yang, Jaime Carbonell, Russ R. Salakhutdinov, and Quoc V. Le. 2019. XLNet: General- ized autoregressive pretraining for language understanding. In Proceedings of the International Conference on Advances in Neural Information Processing Systems. 5754-5764.
- Xiang Zhang, Junbo Zhao, and Yann LeCun. 2015. Character-level convolutional networks for text classification. In Proceedings of the International Conference on Advances in Neural Information Processing Systems. 649-657.
- Caleb Ziems, Ymir Vigfusson, and Fred Morstatter. 2020. Aggressive, repetitive, intentional, visible, and imbalanced: Refining representations for cyberbullying classification. arXiv preprint arXiv:2004.01820 (2020). · arXiv
- Authors
- Cheng, L., Guo, R., Silva, Y. N., Hall, D., & Liu, H.
- Affiliations
- Arizona State University (US)
- Venue
- ACM/IMS Transactions on Data Science
- Published
- 2021-04-08
- Volume
- 2
- Issue
- 2
- Pages
- 1-23
- ISSN
- 2577-3224 (ISSN-L)
- 2691-1922
- OpenAlex
- W3156978600
- Cited by
- 43
- References
- 52
- Citations by year
- 2026: 6 · 2025: 10 · 2024: 6 · 2023: 12 · 2022: 4 · 2021: 5
- FWCI
- 2.39
- Type
- journal article
- Language
- en
- Open access
- bronze
- doi.org
- Locations
- ACM/IMS Transactions on Data Science (publishedVersion)
- Funders
- National Science Foundation (1719722 and 2036127)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Hate Speech and Cyberbullying Detection
- Social Sciences › Psychology › Social Psychology › Bullying, Victimization, and Aggression
- Physical Sciences › Computer Science › Information Systems › Spam and Phishing Detection
- Keywords
- cyberbullying detection, temporal patterns, hierarchical attention network, Instagram, word-level attention, adolescents, online aggression
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
