Effective requirement categorization is essential for successful software engineering projects; nonetheless, it presents challenges due to the intricacies and ambiguities inherent in natural language specifications. This work introduces an innovative approach to enhance classification reliability and transparency by integrating interpretable machine learning algorithms with relationship parsing. In contrast to black-box models, our technique elucidates the rationale behind specific decisions, enabling developers and stakeholders to have a deeper understanding of the logic behind categorical outcomes. Through dependency analysis, we capture the structural and relational links inside requirement statements, thereby enhancing the feature set used in categorization. The proposed technique improves classification efficiency and connects technical representations with human understanding. A thorough evaluation of several datasets demonstrates the use of this technique for software requirements development.