Synthetic corpus offers detailed phases of online grooming conversations
LURE: a Phase-Labelled Synthetic Corpus of Online-Grooming Exchanges
Artificial Intelligence
Summary
Online grooming, where adults try to manipulate minors, happens through conversations that change over time in clear stages. The authors created a large set of fake but realistic conversations showing these stages, using a special language-processing method combined with human review. This dataset helps understand how grooming builds up and can train tools to spot it early. It matches many features of real conversations and includes emotional tags to better capture how the exchanges feel. The dataset and its code are available to support research and tool development.
What this means in practice
- •For online safety teams: Train and test early warning systems to detect grooming phases in chat using the phase-labelled synthetic conversations.
- •For content moderation platforms: Develop and improve automated filters that recognize grooming language patterns and emotional cues in online messages.
Authors
Emanuele Florindi, Valentina Franzoni, Ludovico Guercio, Alfredo Milani, Mattia Polticchia
Abstract
Phase-annotated conversational data is essential for studying the progression of online grooming of minors and for developing early warning tools. However, the only publicly available source (i.e., Perverted-Justice archive, which contains a subset coded using ChatCoder) is small, outdated, and frozen. We present a synthetic corpus of groomer-victim exchanges, generated using a retrieval-augmented pipeline running on institutional hardware. This pipeline uses an openly available, eight-billion-parameter, refusal-ablated language model. The corpus was created using the five stages of the responsible-generation protocol for synthetic datasets: ontology-driven design based on the ChatCoder luring-communication scheme (eight communicative tags mapped onto four grooming phases); multi-model prompt testing; human-in-the-loop supervision; semantic and feature-level validation; and modality-aware documentation. The released corpus contains 2,121 unique exchanges across seven tags and four phases, each accompanied by a Plutchik emotion field. Validation quantifies the integrity and lexical profile of the corpus, as well as the reproduction of real seed lines, in-domain label recoverability (tag accuracy: 0.85; phase accuracy: 0.93) and its transfer to real ChatCoder offender lines, which were coded automatically at the eight-tag level (31,640 lines) and by human coders at the three-construct level (10,876 posts). Accuracy was measured for lexical and sentence-embedding features, while a real-versus-synthetic discriminator was able to separate the two domains with 91% accuracy. The scope of the corpus includes controlled, phase-conditioned experimentation, pipeline development and augmentation alongside real data and awareness material. The corpus, its datasheet and the audit code are released with vetted access.