Year: 2025
Type: conference-paper
Abstract: This paper presents our system for SubTask 1: Islamic Inheritance Reasoning in the QIAS 2025 Shared Task, which evaluates large language models (LLMs) on (ilm al-mawārīth) (Islamic science of inheritance) using a benchmark of Arabic multiple-choice questions (MCQs) derived from expert-reviewed fatwas.We explore static and dynamic fewshot prompting, retrieval-augmented generation (RAG) with a large fatwa corpus, and a progressive n-gram overlap retrieval method.The n-gram method is applied both to select the top five most similar MCQs for dynamic prompting and to retrieve the most relevant fatwa answer as additional context during inference.We evaluate proprietary and opensource LLMs individually and in ensemble form.Results show that dynamic prompting and RAG consistently improve accuracy across our best performing model, Gemini, achieving 62.26% accuracy on the test set.
Author: Shatha Altammami
Institution: King Saud University
Language: English
Topic: Topic Modeling
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4416037620