Penerapan Model Bahasa Qwen Berbasis Retrieval-Augmented Generation (RAG) untuk Solusi Question Answering Otomatis pada Layanan Helpdesk

Al Ghozali, Alif Ridhwan (2026) Penerapan Model Bahasa Qwen Berbasis Retrieval-Augmented Generation (RAG) untuk Solusi Question Answering Otomatis pada Layanan Helpdesk. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5024221034-Undergraduate_Thesis.pdf Restricted to Repository staff only Download (4MB) | Request a copy

Abstract

Layanan helpdesk pada institusi akademik menuntut penyampaian informasi yang responsif, akurat, dan konsisten. Pemanfaatan teknologi Artificial Intelligence (AI), khususnya Large Language Model (LLM), menawarkan potensi efisiensi yang besar, tetapi sering kali terkendala oleh masalah halusinasi ketika model menghasilkan jawaban yang tidak didasarkan pada fakta. Penelitian ini bertujuan mengatasi permasalahan tersebut dengan menerapkan model bahasa Qwen berbasis arsitektur Retrieval-Augmented Generation (RAG) sebagai solusi Question Answering otomatis pada layanan helpdesk Departemen Teknik Komputer ITS. Metode yang diusulkan menggunakan arsitektur Dual-Mode System yang dilengkapi dengan fitur Automatic Routing Detection. Fitur ini secara cerdas mengklasifikasikan intensi pengguna untuk membedakan antara pertanyaan spesifik domain yang memerlukan verifikasi data dan percakapan umum. Sistem mengintegrasikan model Qwen 3 (1.7B) yang dijalankan secara lokal melalui Ollama dengan basis pengetahuan (*knowledge base*) yang berisi dokumen resmi akademik. Melalui pendekatan ini, sistem melakukan proses *retrieval* terhadap dokumen yang relevan sebelum menghasilkan jawaban, sehingga setiap respons memiliki landasan referensi yang valid. Hasil penelitian ini diharapkan dapat membuktikan bahwa pendekatan RAG mampu meningkatkan akurasi jawaban secara signifikan dan meminimalkan risiko halusinasi dibandingkan penggunaan model bahasa secara langsung, sekaligus menyediakan layanan dukungan pengguna yang andal dan mampu beroperasi selama 24 jam.
=======================================================================================================================================
Helpdesk services in academic institutions require responsive, accurate, and consistent information delivery. The adoption of Artificial Intelligence (AI), particularly Large Language Models (LLMs), offers significant potential for improving operational efficiency but is often limited by hallucination, in which the model generates responses that are not grounded in factual information. This study aims to address this challenge by implementing the Qwen language model based on the Retrieval-Augmented Generation (RAG) architecture as an automated Question Answering solution for the Department of Computer Engineering ITS helpdesk service. The proposed method employs a Dual-Mode System architecture equipped with an Automatic Routing Detection feature. This feature intelligently classifies user intent to distinguish between domain-specific inquiries that require knowledge verification and general conversational interactions. The system integrates the Qwen 3 (1.7B) model, deployed locally through Ollama, with a knowledge base containing official academic documents. Using this approach, the system retrieves relevant documents before generating responses, ensuring that every answer is supported by valid reference sources. The results of this study are expected to demonstrate that the RAG approach significantly improves answer accuracy and minimizes hallucination compared with the direct use of a language model, while providing a reliable helpdesk service capable of operating continuously on a 24-hour basis.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Chatbot, Qwen, Large Language Model, Retrieval-Augmented Generation, Chatbot, Qwen, Large Language Model, Retrieval-Augmented Generation
Subjects:	T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T58.5 Information technology. IT--Auditing T Technology > T Technology (General) > T58.6 Management information systems T Technology > T Technology (General) > T58.64 Information resources management T Technology > T Technology (General) > T58.8 Productivity. Efficiency
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Computer Engineering > 90243-(S1) Undergraduate Thesis
Depositing User:	Alif Ridhwan Al Ghozali
Date Deposited:	30 Jul 2026 04:35
Last Modified:	30 Jul 2026 04:35
URI:	http://repository.its.ac.id/id/eprint/139573

Actions (login required)

View Item