Saved in:
Bibliographic Details
Main Authors: Alhazmi, Elaf, Sheng, Quan Z., Zhang, Wei Emma
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.17574
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910146493415424
author Alhazmi, Elaf
Sheng, Quan Z.
Zhang, Wei Emma
author_facet Alhazmi, Elaf
Sheng, Quan Z.
Zhang, Wei Emma
contents Distractor generation (DG) remains a labor-intensive task that still significantly depends on domain experts. The task focuses on generating plausible yet incorrect options, known as distractors, for multiple-choice questions. A reliable distractor must be contextually relevant to the question and able to mislead examinees through implicit reasoning when identifying the correct answer. While a recent method integrates fine-tuning pre-trained encoder-decoder models with contrastive learning to generate semantically relevant distractors for a given question-answer, it often fails to capture the underlying reasoning process that experts utilize when selecting distractors in benchmarks. In this paper, we explore large language models (LLMs) reasoning for DG through in-context learning with unsupervised semantic retrieval for selecting few-shot examples. We design a rationale-augmented DG framework that jointly generates distractors and their rationales for a given question-answer. Extensive experiments on six benchmarks, with varying average distractor lengths and domains, demonstrate that prompting LLMs with few-shot examples substantially improves the performance compared to recent DG models. It outperforms recent approaches and achieves state-of-the-art results in generating reasoned distractors that align with human-labeled benchmarks.
format Preprint
id arxiv_https___arxiv_org_abs_2604_17574
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Beyond Fine-Tuning: In-Context Learning and Chain-of-Thought for Reasoned Distractor Generation
Alhazmi, Elaf
Sheng, Quan Z.
Zhang, Wei Emma
Computation and Language
Distractor generation (DG) remains a labor-intensive task that still significantly depends on domain experts. The task focuses on generating plausible yet incorrect options, known as distractors, for multiple-choice questions. A reliable distractor must be contextually relevant to the question and able to mislead examinees through implicit reasoning when identifying the correct answer. While a recent method integrates fine-tuning pre-trained encoder-decoder models with contrastive learning to generate semantically relevant distractors for a given question-answer, it often fails to capture the underlying reasoning process that experts utilize when selecting distractors in benchmarks. In this paper, we explore large language models (LLMs) reasoning for DG through in-context learning with unsupervised semantic retrieval for selecting few-shot examples. We design a rationale-augmented DG framework that jointly generates distractors and their rationales for a given question-answer. Extensive experiments on six benchmarks, with varying average distractor lengths and domains, demonstrate that prompting LLMs with few-shot examples substantially improves the performance compared to recent DG models. It outperforms recent approaches and achieves state-of-the-art results in generating reasoned distractors that align with human-labeled benchmarks.
title Beyond Fine-Tuning: In-Context Learning and Chain-of-Thought for Reasoned Distractor Generation
topic Computation and Language
url https://arxiv.org/abs/2604.17574