Analyzing User Descriptions of Child Sexual Abuse Material Posted on Darknet Forums: A Manual and Automated Content Analysis
Rebecca Reichel, Meike de Boer, Frederic M. Gnielka, Anton Daser, Arjan Blokland, Alexander F. Schmidt, Katarzyna Staciwa, Robert J. B. Lehmann · Victims & Offenders · 2026
When analyzing darknet child sexual abuse material (CSAM) forum data, the content of the material exchanged through the forum is highly relevant to understand trends and user preferences. The present study used 129,410 posts containing hyperlinks from six large darknet CSAM forums that were seized by the police between 2011 and 2018 to deduce the content of the uploaded CSAM. Adapting the Universal Classification Schema from INHOPE (version 2) to describe the visual content of CSAM, a sample of descriptions accompanying posted hyperlinks were manually coded to create a dictionary of indicative words. In the next step, this dictionary was used to automatically classify all posts according to the Universal Classification Schema. Acceptable interrater reliability between human and automatic classification was found for ten categories (e.g. victim sex and age), which were then used to describe the CSAM posted. Further endorsing the usefulness of automatically classifying posts using the Universal Classification Schema, the current findings concur with previous research in that we infer a higher number of female victims, a peak in victim age around 11–14 years, and the co-occurrence of male victim sex and penetrative sexual activity. Content descriptions, however, did not predict the popularity of posts.