Abstract
Multi-label Text Classification (MLTC) as a problem is a scenario in which a text document can belong to one or more classes simultaneously. Such classification tasks pose several general as well as specific research challenges. The general challenges include dependency among classes, imbalanced data, and scalability in the presence of an excessive number of labels. On the other hand, the MLTC-specific challenges include high dimensional feature space, obtaining contextual and semantic knowledge from the text, and understanding content diversity. This paper provides a brief description of the multi-label classification approaches such as problem transformation, algorithm adaptation, and ensemble learning along with their strengths and weaknesses. Furthermore, we proposed an MLTC framework referred to as HMTCS (Hybrid feature-based Multi-label Text Classification System) that handles both general multi-labeling issues and text categorization-specific issues. The proposed framework has three modules, namely, Labels Knowledge Base, Hybrid Feature Extraction, and Ensemble Learning.