Automated Detection of Cyberbullying and Hate Speech to Support Digital Mental Health Surveillance: A Systematic Literature Review of Machine Learning and Deep Learning Approaches
Al Makki Health Informatics Journal · 25.03.2026 · 4 (2)
DOI: 10.57185/z73dmf32 ↗Kurz gefasst
Systematische Übersicht in der Zeitschrift Al Makki Health Informatics Journal, online erschienen am 25.03.2026. Verfasst von Fitri Nurjanah und 1 weiteren Autor:innen.
Abstract (Originalsprache)
Cyberbullying and hate speech on social media are mental health risk factors, so automated detection is a candidate component of digital mental health surveillance. This systematic literature review, guided by PRISMA 2020, synthesized machine learning and deep learning classification methods and assessed their readiness for mental health screening. Scopus, IEEE Xplore, PeerJ, Frontiers, SpringerLink, and SINTA indexed journals were searched for articles published from 2021 to 2026. Of 482 records, 412 were screened, and 15 studies scoring at least 3.5 of 5.0 on quality assessment were retained: 10 classification studies, which formed the primary evidence, and 5 supporting studies (two reviews, one psychosocial survey, and two general sentiment analyses). Twitter/X was the dominant source (7 of 10 classification studies), and six of ten corpora were Indonesian. Within-study evidence for TF-IDF over Count Vectorizer was limited to one verifiable comparison, with a margin of 1.63 accuracy points. SVM reached 80% to 92% accuracy (85% to 91.5% on Indonesian corpora) and exceeded Naive Bayes in four studies, whereas Naive Bayes ranged from 71.2% to 95% and failed on imbalanced data. A hybrid CNN-BiLSTM reached 99% accuracy with 74.67% recall. Only five of ten studies reported precision, recall, and F1-score together, and two addressed class imbalance. Most studies classified single posts, and none measured health outcomes. Future work should report recall and per-class F1-score on realistic data, model conversation sessions, and keep human review in the loop.
Zitieren
APA
Nurjanah, F., & Nurhakim, B. (2026). Automated Detection of Cyberbullying and Hate Speech to Support Digital Mental Health Surveillance: A Systematic Literature Review of Machine Learning and Deep Learning Approaches. Al Makki Health Informatics Journal, 4(2). https://doi.org/10.57185/z73dmf32
Vancouver
Nurjanah F, Nurhakim B. Automated Detection of Cyberbullying and Hate Speech to Support Digital Mental Health Surveillance: A Systematic Literature Review of Machine Learning and Deep Learning Approaches. Al Makki Health Informatics Journal. 2026;4(2). doi:10.57185/z73dmf32
BibTeX
@article{Nurjanah2026,
author = {Nurjanah, Fitri and Nurhakim, Bani},
title = {Automated Detection of Cyberbullying and Hate Speech to Support Digital Mental Health Surveillance: A Systematic Literature Review of Machine Learning and Deep Learning Approaches},
journal = {Al Makki Health Informatics Journal},
year = {2026},
volume = {4},
number = {2},
doi = {10.57185/z73dmf32}
}
EUVIMED