Predatory Conversation Detection
Abstract
Providing a safe environment for children in online networks can be challenged by the anonymous nature of the internet. One of the worst forms of cyber-security issues is child grooming, where sexual predators seek contact with minors to abuse them. Various types of organizations such as chat providers and law enforcement are inclined to find online predatory conversations in order to protect the children. This paper proposes a study on predatory conversation detection using Natural Language Processing. We analyzed the different types of features of online grooming data considering various characteristics of online conversation, such as psycho-linguistic patterns. Our experiments with online communication showed an accuracy of 0.98 in automatically classifying the conversations into predatory conversations and non-predatory conversations. Best results are obtained by using linear SVM on 1-gram features when removing stop words as well as by using multinomial Naïve Bayes on 1-gram features when not removing stop words.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
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?
The primary way online predators gather information is by asking many questions.
“Their primary way to gather information is by asking many questions [49].”
From Online grooming detection: A comprehensive survey of child exploitation in chat logs (2023) · cites this paper as [49]
The predator might not talk about sexual topics with the victim during the exclusive stage to avoid losing trust so that he can finally approach the victim in person and execute some harmful action.
“One should consider that the predator might not talk about sexual topics with the victim during this stage to avoid losing trust so that he can finally approach the victim in person and execute some harmful action [49].”
From Online grooming detection: A comprehensive survey of child exploitation in chat logs (2023) · cites this paper as [49]
Most computer science literature focuses on detecting perpetrators' behaviors using machine learning and natural language processing techniques.
“To combat cybergrooming, researchers have developed numerous computational approaches, with most computer science literature focuses on detecting perpetrators' behaviors using machine learning (ML) and natural language processing (NLP) techniques (e.g., [5, 16-19, 3234, 42, 44, 48, 59, 62, 72, 76-78]).”
From From Vulnerable to Resilient: Examining Parent and Teen Perceptions on How to Respond to Unwanted Cybergrooming Advances (2026) · cites this paper as [18]
Cites
Detecting child grooming behaviour patterns on social media
Cano, A. E., Fernandez, M., & Alani, H. (2014) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-13734-6_30 · Cited by 79 · see citing works
“Cause ur special”: Understanding trust and complimenting behaviour in online grooming discourse
Lorenzo-Dus, N., & Izura, C. (2017) · Journal of Pragmatics · doi:10.1016/j.pragma.2017.01.004 · Cited by 71 · see citing works
Flagged first, then by citations
Cited by
Sorted by citations
Razi, A., Kim, S., Alsoubai, A., Stringhini, G., Solorio, T., De Choudhury, M., & Wisniewski, P. J. (2021) · Proceedings of the ACM on Human-Computer Interaction · doi:10.1145/3479609 · Cited by 95 · see citing works
Online grooming detection: A comprehensive survey of child exploitation in chat logs
Borj, P. R., Raja, K., & Bours, P. (2023) · Knowledge-Based Systems · doi:10.1016/j.knosys.2022.110039 · Cited by 38 · see citing works
Self-citationEnsemble Method for Sexual Predators Identification in Online Chats
Fauzi, M. A., & Bours, P. (2020) · 2020 8th International Workshop on Biometrics and Forensics (IWBF) · doi:10.1109/iwbf49977.2020.9107945 · Cited by 34 · see citing works
Self-citationDetecting Online Grooming By Simple Contrastive Chat Embeddings
Rezaee Borj, P., Raja, K., & Bours, P. (2023) · Proceedings of the 9th ACM International Workshop on Security and Privacy Analytics · doi:10.1145/3579987.3586564 · Cited by 11 · see citing works
Self-citationDetecting Sexual Predatory Chats by Perturbed Data and Balanced Ensembles
Borj, P. R., Raja, K., & Bours, P. (2021) · 2021 International Conference of the Biometrics Special Interest Group (BIOSIG) · doi:10.1109/biosig52210.2021.9548303 · Cited by 13 · see citing works
Self-citationIdentifying Sexual Predators in Chats Using SVM and Feature Ensemble
Fauzi, M. A., Wolthusen, S., Yang, B., Bours, P., & Yeng, P. (2023) · 2023 International Conference on Emerging Trends in Networks and Computer Communications (ETNCC) · doi:10.1109/etncc59188.2023.10284950 · Cited by 4 · see citing works
Self-citationZhang, X., Akter, M., An, H., Liu, M., Zhang, Q., Huang, L., Cho, J.-H., J. Wisniewski, P., & Lee, S. W. (2026) · Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3772318.3790949
- Authors
- Borj, P. R., & Bours, P.
- Affiliations
- Norwegian University of Science and Technology (NO)
- Venue
- 2019 International Conference on Cyber Security for Emerging Technologies (CSET)
- Publisher
- IEEE
- Published
- 2019-10
- Pages
- 1-6
- OpenAlex
- W2990167730
- Cited by
- 19
- References
- 2
- Citations by year
- 2026: 1 · 2025: 4 · 2024: 2 · 2023: 4 · 2022: 2 · 2021: 4 · 2020: 2
- FWCI
- 0.77
- Type
- conference paper
- Language
- en
- Open access
- closed
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Hate Speech and Cyberbullying Detection
- Physical Sciences › Computer Science › Artificial Intelligence › Text and Document Classification Technologies
- Physical Sciences › Computer Science › Information Systems › Spam and Phishing Detection
- Keywords
- online child grooming, child sexual abuse, text classification, stop word removal, linear support vector machine, multinomial naive Bayes
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
