Implementasi Chatbot Berbasis Artificial Intelligence Sebagai Layanan Informasi Perpustakaan SMA Negeri 2 Sipora Berbasis Web

Fadhlin, Aura (2026) Implementasi Chatbot Berbasis Artificial Intelligence Sebagai Layanan Informasi Perpustakaan SMA Negeri 2 Sipora Berbasis Web. Skripsi thesis, Universitas Islam Negeri Imam Bonjol Padang.

	Text (COVER - DAFTAR ISI) AURA FADHLIN - 2217020063 COVER-DAFTAR ISI.pdf - Published Version Download (1MB)
	Text (BAB 1) AURA FADHLIN - 2217020063 BAB 1.pdf - Published Version Download (310kB)
	Text (BAB 3) AURA FADHLIN - 2217020063 BAB 3.pdf - Published Version Download (416kB)
	Text (BAB 5 - DAFTAR PUSTAKA) AURA FADHLIN - 2217020063 BAB 5 - DAFTAR PUSTAKA.pdf - Published Version Download (326kB)
	Text (SKRIPSI FULL TEXT) AURA FADHLIN - 2217020063 SELESAI.pdf - Published Version Restricted to Repository staff only Download (3MB)

Abstract

The Library of SMA Negeri 2 Sipora still faces challenges in providing information services because some service processes are carried out manually. Users need to ask the librarian directly to obtain information about operating hours, borrowing regulations, and book collections. This condition indicates the need for an information service that can help users access library information more easily and independently. This study aims to implement the Retrieval-Augmented Generation (RAG) architecture in an Artificial Intelligence-based chatbot as a web-based information service for the Library of SMA Negeri 2 Sipora. RAG connects a Large Language Model (LLM) with the library's knowledge base so that relevant information can be retrieved before being used to generate responses. The system was developed using n8n as a workflow automation tool, ChromaDB as a vector database, Ollama to run the model locally, and Gemma3:4b as the LLM. Testing was conducted using Functional Testing, Performance Testing, User Acceptance Testing (UAT), and BERTScore. The Functional Testing results showed a success rate of 100%, Performance Testing obtained an average response time of 90.95 seconds, and UAT achieved a score of 88.8%, categorized as very feasible. BERTScore evaluation of 10 questions resulted in an average Precision of 0.8542, Recall of 0.8476, and F1-Score of 0.8507. These results indicate that the chatbot can perform its designed functions, is well accepted by users, and generates responses with good semantic similarity to the reference answers. Therefore, the implementation of RAG in the chatbot can be used as a web-based information service for the Library of SMA Negeri 2 Sipora.

Actions (login required)

View Item