Sistem Informasi Manajemen Dasbor File Digital Pencarian File Berbasis Kecerdasan Buatan

Kusumah, Mulki (2026) Sistem Informasi Manajemen Dasbor File Digital Pencarian File Berbasis Kecerdasan Buatan. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 05311940000043-Undergraduate_Thesis.pdf Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Penelitian ini mengembangkan sebuah sistem manajemen file berbasis web yang dilengkapi dengan kemampuan pencarian dokumen menggunakan metode Retrieval-Augmented Generation (RAG). Sistem dirancang untuk mempermudah pengguna dalam mengelola, menyimpan, serta mencari dokumen secara lebih cepat dan relevan melalui pemanfaatan embedding vektor dan model Large Language Model (LLM). Pengembangan RAG meliputi proses ekstraksi konten dokumen, chunking, embedding menggunakan model BGE-M3, penyimpanan embedding dalam ChromaDB, serta penggunaan retriever dengan nilai top-k dinamis. Sistem juga disertai API berbasis FastAPI untuk proses indexing dan penjawaban pertanyaan, serta antarmuka web Laravel untuk manajemen file. Pengujian dilakukan melalui tiga kategori: uji performa RAG (Self-Consistency 0.8533 dan MAP 0.7257), uji API (response time unggah file berdasarkan ukuran dan waktu respons pertanyaan), serta uji pengguna menggunakan skala Likert terhadap lima aspek utama. Hasil evaluasi menunjukkan bahwa sistem bekerja dengan baik, memberikan jawaban yang relevan, memiliki antarmuka yang mudah digunakan, serta layak untuk diimplementasikan. Dengan demikian, sistem ini mampu meningkatkan efisiensi pengelolaan dokumen dan kualitas pencarian berbasis kecerdasan buatan.
================================================================================================================================
This research presents the development of a web-based file management system equipped with Retrieval-Augmented Generation (RAG) to support semantic document search. The system is designed to help users manage, store, and retrieve documents more efficiently by leveraging vector embeddings and large language models. The RAG pipeline includes document extraction, dynamic chunking, embedding with the BGE-M3 model, vector storage using ChromaDB, and a retriever configured with dynamic top-k selection. The system also integrates a FastAPI backend for indexing and question-answering, as well as a Laravel web interface for file management. Three categories of evaluation were conducted: RAG performance (Self-Consistency 0.8533 and MAP 0.7257), API performance (file upload response times and question-answering latency), and user evaluation using a Likert-scale questionnaire across five criteria. The results indicate that the system performs effectively, produces relevant answers, offers a user-friendly interface, and is feasible for practical implementation. Overall, the system improves document management efficiency and enhances AI-based document retrieval quality.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Kecerdasan Buatan, Large Language Model (LLM), Retrieval-Augmented Generation (RAG), Sistem Manajemen File, Pencarian Dokumen, Artificial Intelligence, File management system, Document Search, Retrieval-Augmented Generation (RAG), Large Language Model (LLM)
Subjects:	T Technology > T Technology (General)
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information Technology > 59201-(S1) Undergraduate Thesis
Depositing User:	Mulki Kusumah
Date Deposited:	05 Feb 2026 09:44
Last Modified:	05 Feb 2026 09:44
URI:	http://repository.its.ac.id/id/eprint/132199

Actions (login required)

View Item