Multiclass Social Media Dataset for Mental Health and Emotion Classification

This dataset comprises 42,518 unique, fully cleaned text records tailored specifically for mental health detection and emotion classification. The data represents social media textual expressions categorized into six precise emotional and psychological states: Positive, Neutral, Anger, Depression, Sadness, and Anxiety. This corpus contains zero duplicate entries and has been thoroughly pre-cleaned, making it a benchmark dataset for training reliable Machine Learning, Deep Learning (BERT, RoBERTa, BiLSTM), and Ensemble meta-classifiers in computational psychology and NLP.