Identifying Sellers of Illicit Narcotics on Soundcloud.com Using Latent Dirichlet Allocation

Previously, I developed a framework for identifying sellers of illicit narcotics advertising on Soundcloud.com. This framework scraped comments and identified comments that were advertising drugs through a simple keyword search. While this framework did well due to the similar structure of the comments, I wanted to try to improve this framework by using Latent Dirichlet Allocation. Latent Dirichlet Allocation (LDA) is a natural language processing (NLP) model for learning abstract topics of text, also known as topic modeling. LDA will cluster documents into topics, allowing us to classify comments and find comments that are advertising the sale of narcotics.