Controlling Cloze-test Question Item Difficulty with PLM-based Surrogate Models for IRT Assessment
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866909126081118208 |
|---|---|
| author | Zhang, Jingshen Xie, Jiajun Qiu, Xinying |
| author_facet | Zhang, Jingshen Xie, Jiajun Qiu, Xinying |
| contents | Item difficulty plays a crucial role in adaptive testing. However, few works have focused on generating questions of varying difficulty levels, especially for multiple-choice (MC) cloze tests. We propose training pre-trained language models (PLMs) as surrogate models to enable item response theory (IRT) assessment, avoiding the need for human test subjects. We also propose two strategies to control the difficulty levels of both the gaps and the distractors using ranking rules to reduce invalid distractors. Experimentation on a benchmark dataset demonstrates that our proposed framework and methods can effectively control and evaluate the difficulty levels of MC cloze tests. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2403_01456 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Controlling Cloze-test Question Item Difficulty with PLM-based Surrogate Models for IRT Assessment Zhang, Jingshen Xie, Jiajun Qiu, Xinying Computation and Language Artificial Intelligence Computers and Society Item difficulty plays a crucial role in adaptive testing. However, few works have focused on generating questions of varying difficulty levels, especially for multiple-choice (MC) cloze tests. We propose training pre-trained language models (PLMs) as surrogate models to enable item response theory (IRT) assessment, avoiding the need for human test subjects. We also propose two strategies to control the difficulty levels of both the gaps and the distractors using ranking rules to reduce invalid distractors. Experimentation on a benchmark dataset demonstrates that our proposed framework and methods can effectively control and evaluate the difficulty levels of MC cloze tests. |
| title | Controlling Cloze-test Question Item Difficulty with PLM-based Surrogate Models for IRT Assessment |
| topic | Computation and Language Artificial Intelligence Computers and Society |
| url | https://arxiv.org/abs/2403.01456 |