PENERAPAN TEKNOLOGI LARGE LANGUAGE MODEL VISUAL DALAM EVALUASI DAN PERBAIKAN ANTARMUKA PENGGUNA SITUS WEB

    Cherry Adya Zahra, - and Mochamad Iqbal Ardimansyah, - and Yulia Retnowati, - (2025) PENERAPAN TEKNOLOGI LARGE LANGUAGE MODEL VISUAL DALAM EVALUASI DAN PERBAIKAN ANTARMUKA PENGGUNA SITUS WEB. S1 thesis, Universitas Pendidikan Indonesia.

    Abstract

    Kualitas antarmuka pengguna berperan penting dalam menentukan kenyamanan dan efektivitas penggunaan website. Namun, evaluasi UI yang umumnya dilakukan secara manual masih memerlukan waktu lama dan rentan subjektivitas. Oleh karena itu, diperlukan metode evaluasi yang lebih cepat, objektif, dan adaptif terhadap kebutuhan pengguna modern. Penelitian ini berfokus pada penerapan Model Bahasa Besar Multimodal Visual untuk meningkatkan evaluasi Antarmuka Pengguna (UI) dalam desain situs web, khususnya website Hirokui milik GreatEdu. Penelitian ini bertujuan untuk mengidentifikasi kekurangan visual dan tata letak menggunakan model Gemini serta menganalisis dampaknya terhadap kualitas UI berdasarkan metrik kegunaan yang diukur melalui System Usability Scale (SUS). Penelitian menggunakan kerangka Design Research Methodology (DRM) yang mencakup tahapan klarifikasi, studi deskriptif, studi preskriptif, dan evaluasi. Aplikasi yang dikembangkan dengan Flask Python dan Gemini API Vision memungkinkan analisis UI otomatis melalui pemrosesan gambar desain situs web dan menghasilkan rekomendasi yang dapat ditindaklanjuti. Pengujian dilakukan pada karyawan GreatEdu berusia 21–30 tahun dengan jumlah 30 responden, menghasilkan skor SUS rata-rata 82,5 yang menunjukkan tingkat kegunaan tinggi (“Baik” hingga “Sangat Baik”). Temuan menunjukkan bahwa model ini efektif mengidentifikasi 85% masalah UI yang juga ditemukan dalam evaluasi heuristik konvensional, serta mampu memangkas waktu evaluasi dari 20 menit menjadi hanya 5 menit per desain. Meskipun demikian, model masih terbatas dalam menangani faktor kontekstual seperti preferensi budaya. Penelitian ini berkontribusi pada literatur evaluasi UI berbasis kecerdasan buatan sekaligus memberikan panduan praktis bagi desainer untuk meningkatkan kegunaan website. ------ The quality of a User Interface (UI) plays a crucial role in determining the comfort and effectiveness of website usage. However, UI evaluations that are typically conducted manually often require significant time and are prone to subjectivity. Therefore, there is an urgent need for evaluation methods that are faster, more objective, and adaptive to modern user needs. This study focuses on the application of a Visual Multimodal Large Language Model to improve the evaluation of User Interface (UI) in website design, specifically the Hirokui website owned by GreatEdu. The aim of this research is to identify visual and layout deficiencies using the Gemini model and to analyze their impact on UI quality based on usability metrics measured through the System Usability Scale (SUS). The research employs the Design Research Methodology (DRM), consisting of clarification, descriptive study, prescriptive study, and evaluation stages. The developed application, built with Flask Python and Gemini API Vision, enables automatic UI analysis by processing website design images and providing actionable recommendations. Testing was conducted with 30 GreatEdu employees aged 21–30 years, resulting in an average SUS score of 82.5, indicating a high level of usability (“good” to “excellent”). Findings show that the model effectively identified 85% of UI issues also found in conventional heuristic evaluations and significantly reduced evaluation time from 20 minutes to just 5 minutes per design. Nevertheless, the model remains limited in addressing contextual factors such as cultural preferences. This study contributes to the literature on AI-based UI evaluation and provides practical guidance for designers to enhance website usability.

    Baca Full Text klik disini
    [thumbnail of S_RPL_2107690_Title.pdf] Text
    S_RPL_2107690_Title.pdf
    Download (1MB)
    [thumbnail of S_RPL_2107690_Chapter1.pdf] Text
    S_RPL_2107690_Chapter1.pdf
    Download (279kB)
    [thumbnail of S_RPL_2107690_Chapter2.pdf] Text
    S_RPL_2107690_Chapter2.pdf
    Restricted to Staf Perpustakaan
    Download (375kB)
    [thumbnail of S_RPL_2107690_Chapter3.pdf] Text
    S_RPL_2107690_Chapter3.pdf
    Download (442kB)
    [thumbnail of S_RPL_2107690_Chapter4.pdf] Text
    S_RPL_2107690_Chapter4.pdf
    Restricted to Staf Perpustakaan
    Download (572kB)
    [thumbnail of S_RPL_2107690_Chapter5.pdf] Text
    S_RPL_2107690_Chapter5.pdf
    Download (261kB)
    [thumbnail of S_RPL_2107690_Appendix.pdf] Text
    S_RPL_2107690_Appendix.pdf
    Restricted to Staf Perpustakaan
    Download (2MB)
    Official URL: https://repository.upi.edu/
    Item Type: Thesis (S1)
    Additional Information: https://scholar.google.com/citations?user=tloDeygAAAAJ&hl=en ID SINTA Dosen Pembimbing: Mochamad Iqbal Ardimansyah: 6658552 Yulia Retnowati: 6852573
    Uncontrolled Keywords: Model Bahasa Besar Multimodal Visual, Antarmuka Pengguna, Website Hirokui, System Usability Scale, Kecerdasan Buatan, Visual Large Language Model, User Interface, Hirokui Website, System Usability Scale, Artificial Intelligence
    Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
    Q Science > QA Mathematics > QA76 Computer software
    T Technology > T Technology (General)
    Divisions: UPI Kampus cibiru > S1 Rekayasa Perangkaat Lunak
    Depositing User: Cherry Adya Zahra
    Date Deposited: 24 Sep 2025 08:09
    Last Modified: 24 Sep 2025 08:09
    URI: http://repository.upi.edu/id/eprint/137839

    Actions (login required)

    View Item View Item