Few-Shot Prompting for Extractive Quranic QA with Instruction-Tuned LLMs
This addresses the problem of low-resource, semantically rich QA for researchers and practitioners in Arabic NLP, though it is incremental as it applies existing prompting methods to a specific domain.
The paper tackled extractive question answering on the Quran by using few-shot prompting with instruction-tuned large language models like Gemini and DeepSeek, achieving a pAP10 score of 0.637 and showing that this approach outperforms traditional fine-tuned models.
This paper presents two effective approaches for Extractive Question Answering (QA) on the Quran. It addresses challenges related to complex language, unique terminology, and deep meaning in the text. The second uses few-shot prompting with instruction-tuned large language models such as Gemini and DeepSeek. A specialized Arabic prompt framework is developed for span extraction. A strong post-processing system integrates subword alignment, overlap suppression, and semantic filtering. This improves precision and reduces hallucinations. Evaluations show that large language models with Arabic instructions outperform traditional fine-tuned models. The best configuration achieves a pAP10 score of 0.637. The results confirm that prompt-based instruction tuning is effective for low-resource, semantically rich QA tasks.