Lune

ACL Findings2026

Candidate-Aware Retrieval and Reranking for Multiple-Choice Question Answering: Arabic as a Case Study

Yassine Bouziane, Youness Moukafih, Mounir Ghogho

摘要

Large language models (LLMs) have recently achieved impressive results on multiple-choice question answering (MCQA), with retrievalaugmented generation (RAG) emerging as an effective strategy for improving the performance of smaller models. However, existing RAG formulations face persistent challenges: retrieving too many passages often introduces noise, and even when relevant content is retrieved, models may still struggle with partially relevant or conflicting information. Moreover, while LLMs perform strongly on English benchmarks, their accuracy declines substantially on Arabic multi-task evaluations, revealing ongoing issues in cross-lingual transfer and domain adaptation. In this paper, we propose a novel approach, using Arabic as a representative case study, that jointly models the relevance of both the question and its candidate answers when selecting contextual passages. The method employs a lightweight reranker trained with a hybrid regression-triplet loss objective to identify passages that provide discriminative and reliable evidence. Extensive experiments across multiple model sizes and humanities domains show that our approach consistently outperforms both standard RAG baselines and reranker baselines, delivering two-to threefold improvements while remaining competitive with considerably larger models.

问 Lune

几分钟读透这篇 Paper,不用再花一下午。

Lune 能读完全文,也能读引用它的顶会 Paper,讲清它的结论、证据和局限,每一点都注明出自哪篇 Paper。不用硬啃,也不靠猜。

可以从这些问题问起

Lune 可以查阅

全文 · 相关的 Paper

获取答案

免费开始,无需绑卡

脉络

  • ArabicMMLU: Assessing Massive Multitask Language Understanding in Arabic 引用了 Crosslingual Generalization through Multitask Finetuning
  • RankRAG: Unifying Context Ranking with Retrieval-Augmented Generation in LLMs 引用了 Dense Passage Retrieval for Open-Domain Question Answering
它引用的引用它的从左到右按年份排列,引用越多,位置越高。

它引用的 Paper5

  1. RankRAG: Unifying Context Ranking with Retrieval-Augmented Generation in LLMsYue Yu, Wei Ping, Zihan Liu, Boxin Wang 等NeurIPS 2024 · 321 次引用
  2. Crosslingual Generalization through Multitask FinetuningNiklas Muennighoff, Thomas Wang, Lintang Sutawika, Adam Roberts 等ACL 2023 · 319 次引用
  3. Dense Passage Retrieval for Open-Domain Question AnsweringVladimir Karpukhin, Barlas Oguz, Sewon Min, Patrick Lewis 等EMNLP 2020 · 142 次引用
  4. ArabicMMLU: Assessing Massive Multitask Language Understanding in ArabicFajri Koto, Haonan Li, Sara Shatnawi, Jad Doughman 等ACL Findings 2024 · 17 次引用
  5. Syn-QG: Syntactic and Shallow Semantic Rules for Question GenerationKaustubh D. Dhole, Christopher D. ManningACL 2020 · 2 次引用

相关 Paper

  1. Don't Be Misled by Style: A Style-Adaptive Reranker for Capturing Effective Knowledge in Retrieval-Augmented GenerationRuwen Zhang, Bo Liu, Zhang Sheng Xiang, Yida Chen 等ACL 2026
  2. Are Large Language Models Good at Utility Judgments?Hengran Zhang, Ruqing Zhang, Jiafeng Guo, Maarten de Rijke 等SIGIR 2024 · 20 次引用
  3. LongRAG: A Dual-Perspective Retrieval-Augmented Generation Paradigm for Long-Context Question AnsweringQingfei Zhao, Ruobing Wang, Yukuo Cen, Daren Zha 等EMNLP 2024 · 13 次引用
  4. Embedding-Based Context-Aware RerankerYe Yuan, Amin Shabani, Siqi LiuICLR 2026
  5. All Languages Matter: Understanding and Mitigating Language Bias in Multilingual RAGDan Wang, Guozhao Mo, Yafei Shi, Cheng Zhang 等ACL 2026
黄昏的海面,两侧是细线勾勒的悬崖