Pengembangan Chatbot Perusahaan Berbasis Retrieval Augmented Generation dengan Perutean Dinamis untuk Efisiensi Token

Maassobirin, Mukhamad Khafid (2026) Pengembangan Chatbot Perusahaan Berbasis Retrieval Augmented Generation dengan Perutean Dinamis untuk Efisiensi Token. Project Report (Praktek Kerja Lapang dan Magang). Fakultas Ilmu Komputer, Surabaya.

	Text (Cover) LAPORAN PKL COVER.pdf Download (689kB)
	Text (Bab I) LAPORAN PKL BAB 1.pdf Download (136kB)
	Text (Bab II) LAPORAN PKL BAB 2.pdf Restricted to Repository staff only until 1 July 2029. Download (379kB) \| Request a copy
	Text (Bab III) LAPORAN PKL BAB 3.pdf Restricted to Repository staff only until 1 July 2029. Download (2MB) \| Request a copy
	Text (Bab IV) LAPORAN PKL BAB 4.pdf Download (84kB)
	Text (Daftar Pustaka) LAPORAN PKL DAFTAR PUSTAKA.pdf Download (139kB)
	Text (Lampiran) LAPORAN PKL LAMPIRAN.pdf Restricted to Repository staff only until 1 July 2029. Download (2MB) \| Request a copy

Abstract

This study discusses the development of a company chatbot based on Retrieval Augmented Generation (RAG) capable of accurately and efficiently answering user queries by leveraging the company's internal documents. The system integrates several components, including document extraction, hierarchical tree structuring, multi-level summarization, semantic embedding using the BGE-M3 model, and vector storage using a vector database (Qdrant). To improve token consumption efficiency during inference, a dynamic routing mechanism is implemented, directing questions to the most relevant document sections based on intent classification and retrieval results. With this approach, the chatbot can deliver more targeted answers while optimizing resource usage. This system is expected to enhance the efficiency of both internal and external information services within the company.

Item Type:

Monograph (Project Report (Praktek Kerja Lapang dan Magang))

Contributors: