Pengembangan Aplikasi RAGNOZA (Retrieval Augmented Generation Knowledge Analyzer) Legal Document

Oleh :
A. A. Amandias Ngulandara - V3423001 - Sekolah Vokasi

The development of artificial intelligence (AI) technology has encouraged the advancement of the Retrieval-Augmented Generation (RAG) approach, a method that combines information retrieval processes with the generative capabilities of Large Language Models (LLM). In the legal domain, the complexity of documents and the use of formal language can become obstacles for users in obtaining legal information quickly and easily. Therefore, RAGNOZA was developed as an AI-based legal assistant using the RAG approach to support the delivery of legal information in a more understandable manner.RAGNOZA enables users to upload audio files or perform voice recordings, which are then processed through an automatic speech transcription mechanism into text. The generated transcripts are used as input to produce legal materials based on the available knowledge base. The system was developed using the Agile development method by utilizing FastAPI and Streamlit frameworks, PostgreSQL database, embedding technology, vector database, and Large Language Model (LLM). The development results show that the system is capable of managing transcript storage, legal material generation processes, and user history records in a structured manner. The evaluation using the RAGAS framework with the Faithfulness, Answer Relevancy, Context Precision, and Context Recall metrics indicates that the system is capable of generating materials that are aligned with the information contained in the knowledge base and the user's query context. Therefore, RAGNOZA can assist users in obtaining legal information more quickly, effectively, and understandably.