Risk Assessment Model
A text classification model fine-tuned on DistilBERT for automated reputational risk assessment of textual reviews and comments. The model classifies input text into three risk levels: Low, Medium, and High.
Model Details
| Property | Details |
|---|---|
| Base Model | DistilBERT (distilbert-base-uncased) |
| Task | Text Classification |
| Fine-tuning Data | ~1,000 manually labeled reviews |
| Language | English |
Intended Use
This model is designed for automated screening of reviews, comments, and user feedback in the context of reputational risk management. It is intended to assist human analysts rather than replace them in decision-making workflows.
Usage
from transformers import pipeline
pipe = pipeline("text-classification", model="jamal-ibrahim/risk_assesment")
result = pipe("The company has been accused of fraud.")
print(result)
Example output:
[{"label": "High", "score": 0.92}]
Classes
| Label | Description |
|---|---|
| Low | No significant reputational risk detected |
| Medium | Moderate concerns that may warrant further review |
| High | Severe reputational implications requiring immediate attention |
Limitations
- Trained on a relatively small, domain-specific dataset (~1,000 entries)
- Not suitable for legal or financial decision-making without qualified human oversight
- Generalization to domains outside the training distribution is not guaranteed
- Detailed training metrics and hyperparameters are not currently available
License
This model is licensed under the Apache 2.0 License, consistent with the base DistilBERT model.
- Downloads last month
- 107,440
Model tree for jamal-ibrahim/risk_assesment
Base model
distilbert/distilbert-base-uncased