Predicting Cyber-Trafficking Websites Using a Naive Bayes Algorithm, Logistic Regression, KNN, and SVM

Researchers and software developers continue discovering the best approach to combat the rising cyber-trafficking issues. However, most studies focus only on one platform or one of the gateways of cyber-trafficking. Thus, this paper introduces the development and comparison of the Naive Bayes Algorithm, Logistic Regression, k-nearest neighbor (KNN), and Support Vector Machine (SVM) classification models to predict trafficking and non-trafficking websites. In developing the supervised classification models, 37 keywords were used to scrape suspected trafficking websites. Thirty-five (35) websites were classified as trafficker out of 63; this data was used to create the models. Upon evaluating the accuracy rates of the models, the Naive Bayes Algorithm got ninety-one percent (91%), Logistic Regression got eighty-one percent (81%), KNN got sixty-four percent (64%), and SVM got sixty-four percent (64%). Thus, Naive Bayes can predict more accurately than the other classification algorithms. The result shows that the predictive model could be an effective tool for identifying different online platforms that are used in trafficking. Once the model is integrated into an application, this will be easier and faster for law enforcement agencies to monitor human trafficking in a fast-growing cyberspace community.