BanglaDep: A Labeled Dataset for Depression Severity Assessment in Bangla Text

The proposed dataset comprises 10,000 Bangla text samples annotated for depression severity across four categories: non-depressive, mild, moderate, and severe. Each text instance captures varying levels of emotional expression, ranging from neutral or positive to highly depressive content. By providing a balanced distribution of samples across severity levels, this dataset offers a reliable benchmark for building and evaluating NLP models in the domain of mental health assessment. As one of the first large-scale resources for Bangla depression severity analysis, it contributes significantly to low-resource language processing, enabling research on automatic detection, severity classification, and explainable interventions