santa47/cbt-distortion-classifier-v1

CBT Cognitive Distortion Classifier

Classifies text into cognitive distortion groups for CBT therapy applications.

Model Details

• Best Model: SVM
• Model Type: Traditional ML
• Classes: 5 distortion groups

Performance

Metric	Clean Data	Real-World
Accuracy	0.4188	0.3781
F1 Score	0.4181	0.3637

Distortion Groups

| Group | Distortions | Treatment | |-------|-------------|-----------|| | G0 | No Distortion | Supportive listening | | G1 | All-or-nothing, Labeling | Gradient restructuring | | G2 | Overgeneralization, Mind Reading, Fortune-telling | Evidence testing | | G3 | Mental filter, Magnification | Attention rebalancing | | G4 | Emotional Reasoning, Personalization, Should statements | Metacognitive distancing |

All Models Compared (Real-World F1)

Model	Accuracy	F1 Score
SVM	0.3781	0.3637
Logistic Regression	0.3406	0.3256
Random Forest	0.3031	0.2765
XGBoost	0.2656	0.2538
TinyBERT	0.2500	0.1982

Usage

See config.json for model type and loading instructions.

Disclaimer

For educational and supportive purposes only. Not a substitute for professional mental health care.