Please use this identifier to cite or link to this item: http://repository.ipb.ac.id/handle/123456789/153554
Title: Penerapan Analisis Sentimen pada Saran Evaluasi Proses Belajar Mengajar di IPB University Menggunakan IndoBERT dan LDA
Other Titles: Application of Sentiment Analysis on Evaluation of Teaching and Learning Process Suggestion at IPB University Using IndoBERT and LDA
Authors: Asfarian, Auzi
Mushthofa
Safrina, Azwa
Issue Date: 2024
Publisher: IPB University
Abstract: Evaluasi Proses Belajar Mengajar menjadi kunci dalam meningkatkan kualitas pembelajaran dan mendukung perkembangan mahasiswa. Namun, permasalahan dalam evaluasi sering terkait dengan ketidakseimbangan antara hasil belajar dan kualitas proses pembelajaran. Analisis sentimen diharapkan dapat memberikan wawasan mendalam terhadap saran mahasiswa pada proses belajar mengajar dengan menggunakan IndoBERT dan Latent Dirichlet Allocation. Penelitian ini bertujuan mengklasifikasikan analisis sentimen ke dalam kategori positif, negatif, dan netral serta mengidentifikasi topik utama yang sering dibicarakan dan menjadi permasalahan serta relevan pada saran evaluasi proses belajar mengajar. Tahapan dari penelitian ini meliputi pengumpulan data, pra proses data, pelabelan data, pembagian data, pemodelan, evaluasi serta pemodelan topik. Dari penelitian ini didapatkan hasil bahwa model IndoBERT mampu mengklasifikasikan sentimen dengan tingkat akurasi 94% sementara pemodelan topik menggunakan Latent Dirichlet Allocation mampu mengidentifikasi topik utama dari saran-saran mahasiswa pada setiap kategori.
Evaluation of the Teaching and Learning Process is key in improving the quality of learning and supporting student development. However, problems in evaluation are often related to the imbalance between learning outcomes and the quality of the learning process. Sentiment analysis is expected to provide in-depth insight into student suggestions on the teaching and learning process using IndoBERT and Latent Dirichlet Allocation. This research aims to classify sentiment analysis into positive, negative, and neutral categories and identify the main topics that are often discussed and become problems and relevant to the evaluation of the teaching and learning process. The stages of this research include data collection, data pre-processing, data labeling, data sharing, modeling, evaluation and topic modeling. From this research, it is found that the IndoBERT model is able to classify sentiment with 94% accuracy while topic modeling using Latent Dirichlet Allocation is able to identify the main topics of student suggestions in each category.
URI: http://repository.ipb.ac.id/handle/123456789/153554
Appears in Collections:UT - Computer Science

Files in This Item:
File Description SizeFormat 
cover_G6401201013_d32082aac5a94d86b53ce9f0a34ddac6.pdfCover843.4 kBAdobe PDFView/Open
fulltext_G6401201013_2004973f6f9b42aabcb4ce71422635e2.pdf
  Restricted Access
Fulltext1.73 MBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.