This paper presents DebtHunter, a natural language processing (NLP)- and machine learning (ML)- based approach for identifying and classifying SATD in source code comments. The proposed classification approach combines two classification phases for differentiating between the multiple debt types.