Using data gleaned from escort service websites, we inferred the topology of the U.S. sex-trafficking network. This information was further analyzed to identify prostitution hubs across the country.
Each U.S. locale was characterized by the collection of phone numbers contained in its associated escort listings. These sets were used to compute the pairwise similarity between locales. The strongest connections are visualized on a map of the U.S. The association graph was used to cluster states, yielding distinct prostitution rings throughout the country. We further extracted the pricing information from escort listings to model the unique demand signals in each cluster.
This project sheds light on potential arteries of domestic sex trafficking and shows where local law enforcement collaboration could be most fruitful.