CoChat
Sign Up
Cited by 28Hate Speech and Cyberbullying DetectionText and Document Classification TechnologiesAuthorship Attribution and Profiling

Detection of Cyber Grooming in Online Conversation

Bours, Patrick, Kulsrud, Halvor·2019·2019 IEEE International Workshop on Information Forensics and Security (WIFS)
DOI 10.1109/wifs47025.2019.9035090
Open this paper in CoChat — free account

Abstract

In this paper we will focus on the detection of sexual predators in online chat conversations. We use 3 different approaches (message-based, author-based and conversation-based) combined with 5 different classification algorithms and 2 different features sets. The best results were obtained using either the author-based approach with the Neural Network classifier on the TF-IDF feature set, or the conversation-based approach using the Ridge or the Naïve Bayes classifier on the TF-IDF feature set. In this paper, for the first time, we looked at how quick a predator could be detected, and found that in most cases 26-161 messages of a conversation were sufficient. This constitutes only a small fraction of the full conversations, showing that we can have an early detection system of sexual predators instead of knowing in retrospect that a child was the victim of a sexual predator.

Integrity signals

Registry

Crossref · 10.1109/wifs47025.2019.9035090

Cites

16 of 16 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Claims the paper makes

This paper’s full text isn’t openly available, so its claims and self-check weren’t run. The registry and citation checks still apply.

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?

  • Some works have explored various phases of grooming to investigate different themes in online conversations, while others focused on predatory conversation detection, predatory identification, or author profiling.

    “Some works [28,33,40,42,43] have explored various phases of grooming to investigate the different themes in online conversations, while others focused on predatory conversation detection [49][50][51], predatory identification [41,52,53], or author profiling [44][45][46][47][48].”

    From Online grooming detection: A comprehensive survey of child exploitation in chat logs (2023) · cites this paper as [53]

  • Some papers used a beta value higher than one to emphasize recall and aim for a lower number of false negative classifications.

    “Some papers [52,53] used a β value higher than one that would emphasize recall and aim for a lower number of false negative classifications.”

    From Online grooming detection: A comprehensive survey of child exploitation in chat logs (2023) · cites this paper as [53]

  • Some words such as 'sweetie', 'hun', 'mwah', and 'lil slut' indicated predatory conversations strongly.

    “For the message-based model, they used LR and Ridge classifiers on the PAN2012 data and found that some words such as ''sweetie'', ''hun'', ''mwah'', and ''lil slut'' indicated predatory conversations strongly.”

    From Online grooming detection: A comprehensive survey of child exploitation in chat logs (2023) · cites this paper as [53]

  • The best performance for the models was obtained using TF-IDF features and the NB classifier.

    “Various models such as LR, Ridge, NB, SVM, and NN were trained by BoW and TF-IDF features, while the best performance was obtained using TF-IDF features and the NB classifier [53].”

    From Online grooming detection: A comprehensive survey of child exploitation in chat logs (2023) · cites this paper as [53]

  • Few earlier works have tried to identify predators in online chats.

    “Few earlier works mentioned above have also tried to identify the predators [51][52][53]71].”

    From Online grooming detection: A comprehensive survey of child exploitation in chat logs (2023) · cites this paper as [53]

Cites

Flagged first, then by citations

1 of the 17 references couldn’t be matched to a known paper.

Cited by

Sorted by citations

Related

Authors
Bours, P., & Kulsrud, H.
Affiliations
Norwegian University of Science and Technology (NO)
ORCID
Bours, Patrick
Venue
2019 IEEE International Workshop on Information Forensics and Security (WIFS)
Publisher
IEEE
Published
2019-12
Pages
1-6
ISSN
2157-4774 (ISSN-L)
DOI
10.1109/wifs47025.2019.9035090
OpenAlex
W3010709518
Cited by
27
References
17
Citations by year
2026: 2 · 2025: 1 · 2024: 3 · 2023: 9 · 2022: 5 · 2021: 6 · 2020: 1
FWCI
1.15
Type
conference paper
Language
en
Open access
closed
Licence
Locations
IEEE International Workshop on Information Forensics and Security (publishedVersion)
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 › Artificial Intelligence › Authorship Attribution and Profiling
Keywords
online grooming, sexual predators, child sexual abuse, early detection, TF-IDF features, neural network classifier, Naive Bayes classifier, ridge regression

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