Bridging the Justice Gap: An AI-Powered Approach to Algerian Legal Accessibility

Abstract

This paper introduces "Hokoki AI," a pioneering AI-powered mobile application engineered to tackle the critical challenge of legal information accessibility in Algeria. The Algerian legal system, with its blend of civil and Islamic law and a scarcity of modern, user-friendly digital resources, presents significant barriers to legal literacy. Our work addresses this by presenting an innovative solution that simplifies complex legal content through an intelligent, multilingual chatbot and an intuitive legal search system.At the core of the Hokoki AI platform is a powerful technical architecture that leverages transformer-based models, specifically a fine-tuned T5-Small, to generate human-like responses. Crucially, we employ a Retrieval-Augmented Generation (RAG) framework, which significantly enhances the system’s accuracy and relevance by grounding its outputs in a custom-curated dataset of Algerian labor law. This approach not only mitigates the risk of AI "hallucinations" but also yielded a measurable 20% gain in performance metrics such as accuracy, precision, recall, and F1 score, as demonstrated in our evaluation. This paper details the problem, objectives, and methodology, offering a compelling case study on the practical application and impact of advanced AI, such as RAG, in democratizing legal technology within a specific cultural and legal context.

Ähnliche Arbeiten