SynTraffic: A synthetic dataset for spatiotemporal human trafficking detection in heterogeneous graphs

• João Lopes Universidade de São Paulo (USP)
• Ana Nascimento Universidade de São Paulo (USP)
• Pratheeksha Nair McGill University / Mila - Quebec AI Institute
• Catalina Vajiac Carnegie Mellon University (CMU)
• Andreas Olligschlaeger i3 LLC
• Reihaneh Rabbany McGill University / Mila - Quebec AI Institute
• Christos Faloutsos Carnegie Mellon University (CMU)
• Mirela Cazzolato Universidade de São Paulo (USP) https://orcid.org/0000-0002-4364-010X

DOI: https://doi.org/10.5753/sbbd_estendido.2026.249933

Resumo

Online escort platforms have become a common vector for human trafficking (HT) networks to advertise victims alongside legitimate listings, making manual screening infeasible. Progress on automated analysis is limited by restricted data access: HT advertisement corpora carry personal, potentially victim-identifying information, imposing ethical and legal constraints on sharing and reproducibility. This work addresses that gap with a reproducible methodology for generating synthetic HT advertisement data that is statistically representative of a real corpus, without exposing sensitive content. The proposed dataset SynTraffic reproduces posting volume, spatial mobility, temporal dynamics, and metadata-sharing structure, including the size distribution of identity clusters formed by shared contact metadata. We statistically validate SynTraffic testing distributional fidelity and cluster-size fidelity. SynTraffic is openly available, as well as the the generation methodology, presenting a privacy-safe substitute for the real corpus.

Palavras-chave: Human trafficking, heterogeneous data, open dataset, synthetic data generation

Referências

Beduschi, A. (2024). Synthetic data protection: Towards a paradigm change in data regulation? Big Data &; Society, 11(1). DOI: 10.1177/20539517241231277.

International Labour Organization, Walk Free, and International Organization for Migration (2022). Global estimates of modern slavery: Forced labour and forced marriage. Technical report, International Labour Organization, Geneva. Accessed: 21 May 2026.

Kulshrestha, A. (2021). Detection of organized activity in online escort advertisements. Master’s thesis, McGill University, Montreal, Canada. URL: [link].

Lee, M.-C., Vajiac, C., Kulshrestha, A., Levy, S., Park, N., Jones, C., Rabbany, R., and Faloutsos, C. (2021). InfoShield: Generalizable information-theoretic human-trafficking detection. In IEEE ICDE. DOI: 10.1109/ICDE51399.2021.00101.

Liu, Z., Chen, C., Yang, X., Zhou, J., Li, X., and Song, L. (2018). Heterogeneous graph neural networks for malicious account detection. In ACM CIKM, page 2077–2085, New York, NY, USA. DOI: 10.1145/3269206.3272010.

Nagpal, C., Miller, K., Boecking, B., and Dubrawski, A. (2017). An entity resolution approach to isolate instances of human trafficking online. In Proceedings of the 3rd Workshop on Noisy User-generated Text, pages 77–84, Copenhagen, Denmark. Association for Computational Linguistics. DOI: 10.18653/v1/w17-4411.

Nair, P., Liu, J., Vajiac, C., Olligschlaeger, A., Chau, D. H., Cazzolato, M., Jones, C., Faloutsos, C., and Rabbany, R. (2024). T-net: weakly supervised graph learning for combatting human trafficking. AAAI. DOI: 10.1609/aaai.v38i20.30233.

Summers, L., Shallenberger, A. N., Cruz, J., and Fulton, L. V. (2023). A multi-input machine learning approach to classifying sex trafficking from online escort advertisements. Machine Learning and Knowledge Extraction, 5(2):460–472.

Vajiac, C., Chau, D. H., Olligschlaeger, A. M., Mackenzie, R., Nair, P., Lee, M., Li, Y., Park, N., Rabbany, R., and Faloutsos, C. (2023a). Trafficvis: Visualizing organized activity and spatio-temporal patterns for detecting and labeling human trafficking. IEEE Trans. Vis. Comput. Graph., 29(1):53–62. DOI: 10.1109/TVCG.2022.3209403.

Vajiac, C., Lee, M.-C., Kulshrestha, A., Levy, S., Park, N., Olligschlaeger, A., Jones, C., Rabbany, R., and Faloutsos, C. (2023b). Deltashield: Information theory for human-trafficking detection. ACM Trans. Knowl. Discov. Data, 17(2). DOI: 10.1145/3563040.