How data mining can help chase sex traffickers
To aid law enforcement, researchers developed automated data mining techniques that can identify human traffickers by linking bitcoin, online advertisements and writing style.
New automated data linkage and mining techniques could help law enforcement more quickly and efficiently find online sex traffickers by linking data from online advertisements to public information from Bitcoin.
The techniques mine data from online sex ads and virtual currency payments for those services, according to a paper presented at the SIGKDD International Conference on Knowledge Discovery and Data Mining on Aug. 16.
The vast pool of online ads makes manual exploration and analysis a tall task for law enforcement or outside groups. Discerning whether an ad is soliciting for a victim of human trafficking or is an "independent sex worker" isn't easy either, said the research team.
The techniques developed by computer scientists at the University of California at Berkeley, UC San Diego and New York University are designed to discern the true authors of online sex ads based on the writing style. They rely on machine learning that purports to detect unique authorship with 96 percent accuracy. That information is key in identifying the person buying the ad, according to the paper.
The researchers combined that technique with another that picks up leakage from the Bitcoin "mempool" blockchain and sex ad sites and refers the data back to Bitcoin public wallets and transactions to further identify and quantify the ad purchaser.
The group worked for a month using the techniques on the notorious Backpage sex ad site to demonstrate how an analyst could use them to link payments, ads and ad authors.
The researchers said they are currently collaborating with multiple non-governmental organizations that track human trafficking and law enforcement officers to deploy the tools. They said they plan to share the tools and data with those groups, as well as make them publicly available.