Model Card: RoBERTa + Personality + NLI for Suicide Ideation Detection
Model Overview
This model detects suicidal ideation in social media posts while reducing false positives from supportive, awareness, and preventive content.
It extends a standard transformer-based classifier by incorporating:
- Psycholinguistic signals (Big Five personality traits)
- Natural Language Inference (NLI) for intent refinement
The goal is to improve intent-aware suicide ideation detection, not just keyword-based classification.
Model Details
- Base Model: RoBERTa
- Enhancements:
- Big Five personality traits (Neuroticism, Extraversion, Agreeableness)
- NLI-based post-classification filtering
- Input: Social media text (Reddit-style posts)
- Output: Binary classification
- Suicidal Ideation
- Non-suicidal content
Intended Use
This model is designed for research in:
- Suicide ideation detection
- Mental health NLP systems
- False positive reduction in safety-critical classifiers
- Intent-aware text classification
Not intended for clinical diagnosis or autonomous decision-making.
Key Contribution
Most existing datasets label suicide-related content uniformly, leading to misclassification of:
- Supportive posts
- Awareness campaigns
- Preventive discussions
This model addresses that issue through:
- Personality-aware representation learning
- NLI-based intent filtering to distinguish discussion vs personal ideation
Performance Summary
Evaluated on standard and supportive-content false positive benchmarks:
- F1-score: 0.94
- Accuracy: 94%
- False Positive Rate (Supportive Content): 14.61%
Notable improvement over baseline transformer models in reducing misclassification of non-ideation content.
Live Demo
Try the interactive demo here:
👉 https://huggingface.co/spaces/lensy111/roberta_suicide_ideation_personality_nli
Related Resources
- Downloads last month
- 20
Model tree for lensy111/roberta_base_suicide_ideation_big5
Base model
FacebookAI/roberta-base