Chatbot Layanan Informasi Unit Pelayanan Terpadu Institut Teknologi Kalimantan dengan Pendekatan Komparasi pada Large Language Model

Maryono, Adimasdefatra Bimasena Sulthantik (2025) Chatbot Layanan Informasi Unit Pelayanan Terpadu Institut Teknologi Kalimantan dengan Pendekatan Komparasi pada Large Language Model. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5027211040-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2027. Download (4MB) | Request a copy

Abstract

Institut Teknologi Kalimantan (ITK) menghadapi tantangan dalam menyediakan layanan informasi yang cepat dan akurat kepada masyarakat umum dan civitas akademika. Penelitian ini bertujuan mengembangkan chatbot layanan informasi berbasis Komparasi pada Large Language Model (LLM) untuk mengotomatisasi penyampaian informasi terkait layanan akademik, administrasi, dan informasi umum di Unit Pelayanan Terpadu (UPT) ITK. Chatbot dirancang melalui proses fine-tuning model LLM menggunakan dataset yang mencakup Frequently Asked Questions (FAQ) dan panduan layanan spesifik ITK. Pendekatan Komparasi digunakan untuk mempercepat pengembangan model dan meningkatkan akurasi jawaban chatbot. Metode yang diterapkan melibatkan preprocessing data, pembagian data, serta pelatihan pada sumber di web huggingface dan mencari model Text Classification yang mendukung bahasa Indonesia dan paling banyak digunakan. Ketiga model tersebut menggunakan tiga model utama: BERT, DistilBERT, dan RoBERTa. Evaluasi pada saat dilatih menunjukkan bahwa model RoBERTa memberikan hasil terbaik pada akurasinya. Hasil penelitian menunjukkan bahwa chatbot mampu meningkatkan efisiensi penyampaian informasi, sekaligus mengurangi beban kerja administrasi. Implementasi chatbot ini tidak hanya mempercepat proses penyampaian informasi, tetapi juga meningkatkan kualitas layanan di ITK melalui pengelolaan data yang lebih terpusat dan terorganisir.
==============================================================================================================================
Institut Teknologi Kalimantan (ITK) faces challenges in providing fast and accurate information services to the general public and the academic community. This research aims to develop an information service chatbot based on Comparison in Large Language Model (LLM) to automate the delivery of information related to academic services, administration, and general information at the Unit Pelayanan Terpadu (UPT) ITK. The chatbot is designed through a fine-tuning process of the LLM model using a dataset that includes Frequently Asked Questions (FAQ) and ITK-specific service guidelines. The Comparison approach is used to accelerate model development and improve the chatbot’s answer accuracy. The method applied involves preprocessing data, data splitting, and training on resources from the huggingface website, as well as searching for Text Classification models that support the Indonesian language and are widely used. These three main models include BERT, DistilBERT, and RoBERTa. Evaluations during training showed that the RoBERTa model provided the best results in terms of accuracy. The research results show that the chatbot can improve the efficiency of information delivery while reducing administrative workloads. The implementation of this chatbot not only accelerates the process of delivering information but also enhances the quality of services at ITK through more centralized and organized data management.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Chatbot, ITK, Large Language Model,Layanan Informasi, Komparasi, Information Service, Large Language Model, Comparison
Subjects:	T Technology > T Technology (General) T Technology > T Technology (General) > T57.5 Data Processing
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information Technology > 59201-(S1) Undergraduate Thesis
Depositing User:	Adimasdefatra Bimasena Sulthantik Maryono
Date Deposited:	04 Feb 2025 01:56
Last Modified:	04 Feb 2025 01:56
URI:	http://repository.its.ac.id/id/eprint/118019

Actions (login required)

View Item