Detection of Human Trafficking Ads in Twitter Using Natural Language Processing and Image Processing

Abstract

Human trafficking that aims at the sexual exploitation of minors is a problem that affects the world; this crime has evolved with the use of the Internet. To make a contribution that facilitates the work of the Police, we have developed a method that uses Natural Language Processing and Image Processing techniques to detect messages on Twitter related to this felony. If minors are used for sexual exploitation, the Law in most countries, consider them human trafficking victims. The system has two phases to recognize the gender and age group of very young people. In the first one, it captures Twitter messages that are suspicious of being related to the crime through specific normalized hashtags. In the second phase, the system recognizes gender and age groups using facial features and or upper body geometry and proportions using Haar filters and SVM algorithm.