PENERAPAN RETRIEVAL AUGMENTED GENERATION DALAM MEMBANGUN CHATBOT KONSULTASI KULIT BERBASIS WEB

In Indonesia, tropical climate, poor sanitation, and limited access to doctors—particularly in remote areas—are the main causes of skin diseases. Direct consultation with dermatologists is often difficult for many individuals to access. This study aims to develop a skin disease consultation chatbot based on the Retrieval Augmented Generation (RAG) approach by utilizing the Langchain framework, the LLaMA model, and the Qdrant vector database. The research data consists of 30 types of skin diseases obtained from the National Library of Medicine. The preprocessing stage includes space normalization, removal of special characters, and handling of missing values to ensure data consistency prior to vectorization. Evaluation results show high Faithfulness (0.9429) and LLMContextRecall (0.9600) scores, indicating the relevance and alignment of the
answers with the source documents. However, the relatively low Precision score (0.4720) indicates a need for improvement in the accuracy of the information. The chatbot's integration with the Chainlit platform provides an interactive interface
that supports login, conversation history, and user feedback features, enhancing user experience and feedback-based development potential. The system achieves optimal retrieval time (0.08–0.29 seconds) and answer generation time is (3.92–6.17 seconds). Future development suggestions include improving answer accuracy, optimizing the model, enriching the medical reference dataset, and adding automated medical validation features to ensure the reliability of health
consultations. Therefore, this chatbot system is expected to serve as an efficient and cost-effective alternative solution to provide initial information on skin diseases for individuals with limited access to healthcare services.
Keywords: Retrieval Augmented Generation, Langchain, LLaMA, Chatbot, Chainlit