Response-free item difficulty modelling for multiple-choice items with fine-tuned transformers: Component-wise representation and multi-task learning

1. 1Fine-tune a transformer encoder on item wording
2. 2Apply component-wise representation to capture inferential demands
3. 3Use multi-task learning to improve model performance
4. 4Evaluate the model on a dataset of multiple-choice items
5. 5Refine the model by analyzing errors and adjusting hyperparameters

Data scientists and AI engineers on a team can benefit from this approach to improve the accuracy of item difficulty modeling, while educators can use the results to inform their teaching practices

💡 Fine-tuning transformer encoders can capture complex relationships between item wording and difficulty

🤖 Fine-tuned transformers can model item difficulty in multiple-choice questions without relying on response data! 💡