Penerapan Teknologi Langchain dan LLM pada Sistem Question Answering Berbasis Chatbot Telegram: Literature Review

Abstract

This study explores the application of LangChain technology and Large Language Models (LLM) in a Question Answering (QA) system based on Telegram chatbots, designed to provide accurate and relevant automatic answers on various topics. The main objective of this research is to utilize LangChain and LLM to enhance the interactivity and accuracy of the chatbot system in answering questions related to Islamic Business Fiqh, Tafsir Al-Jalalain, Tafsir Al-Azhar, and Health Law. This research investigates how these technologies can be integrated into a Telegram-based chatbot to process complex questions and provide contextually appropriate answers. The system processes data using techniques like chunking and embeddings to match questions with relevant answers effectively. Testing results indicate that the system delivers high accuracy in providing answers, as evaluated through BERTScore and ROUGE Score metrics. Despite this, challenges remain in handling long texts with multiple interpretations and ensuring answers are highly specific to the context of each question. User Acceptance Testing (UAT) showed positive results, with high user satisfaction, but also highlighted the need for improvement in contextual management and reducing information errors or hallucinations. This study recommends further development, including improving chatbot interactivity to respond better to various types of questions and enhancing the system’s ability to provide more natural and contextually relevant answers. Overall, the study demonstrates the significant potential of LangChain and LLM technologies in advancing chatbot systems, making them effective solutions for diverse applications in religious, legal, and other fields.