PENGEMBANGAN ASISTEN VIRTUAL BERBASIS LLM DAN LANGCHAIN UNTUK ANALISIS KOMPARATIF ARTIKEL JURNAL BERBAHASA INGGRIS DENGAN PENYAJIAN RINGKASAN KONTEKSTUAL BERBAHASA INDONESIA

    Nurhikam, - and Liptia Venica, - and Mahmudah Salwa Gianti, - (2025) PENGEMBANGAN ASISTEN VIRTUAL BERBASIS LLM DAN LANGCHAIN UNTUK ANALISIS KOMPARATIF ARTIKEL JURNAL BERBAHASA INGGRIS DENGAN PENYAJIAN RINGKASAN KONTEKSTUAL BERBAHASA INDONESIA. S1 thesis, Universitas Pendidikan Indonesia.

    Abstract

    Keterbatasan akses dan pemahaman terhadap artikel ilmiah berbahasa Inggris menjadi tantangan bagi komunitas akademik di Indonesia, terutama di tengah pertumbuhan pesat jumlah publikasi ilmiah global. Hal ini menghambat proses kajian literatur yang mendalam dan relevan. Penelitian ini bertujuan mengembangkan asisten virtual yang mampu melakukan analisis komparatif otomatis terhadap dua artikel jurnal dan menyajikannya dalam ringkasan kontekstual berbahasa Indonesia. Jenis penelitian ini adalah Research and Development (R&D) dengan pendekatan Waterfall. Sistem dikembangkan menggunakan model LLM GPT-4o-mini yang diorkestrasi oleh LangChain, serta antarmuka Streamlit. Evaluasi dilakukan melalui pengujian performa, validasi oleh tiga ahli, dan pengukuran akurasi semantik menggunakan BERTScore. Hasil menunjukkan waktu pemrosesan rata-rata 3 menit 50 detik. Validasi ahli memberikan skor 4,47 dari 5 (89,31%), dan akurasi semantik mencapai BERTScore 0,80. Usability testing menghasilkan skor 4,15 dari 5 (83,00%), membuktikan sistem mudah digunakan dan akurat. Sistem ini efektif menjembatani kesenjangan bahasa dan kompleksitas analisis, serta mempercepat kajian literatur bagi akademisi di Indonesia. ----- Limited access to and understanding of English-language scientific articles remains a significant challenge for the academic community in Indonesia, especially amidst the rapid growth of global scientific publications. This hampers the ability to conduct in-depth and relevant literature reviews. This study aims to develop a virtual assistant capable of automatically performing comparative analysis between two journal articles and presenting the results in contextual summaries in Indonesian. The research adopts a Research and Development (R&D) approach using the Waterfall model. The system is developed using the GPT-4o-mini large language model orchestrated by LangChain, with a Streamlit-based interface. Evaluation was conducted through performance testing, expert validation by three domain experts, and semantic accuracy measurement using BERTScore. Results show an average processing time of 3 minutes and 50 seconds. Expert validation yielded a quality score of 4.47 out of 5 (89.31%), and semantic accuracy reached a BERTScore of 0.80. Usability testing produced a score of 4.15 out of 5 (83.00%), indicating that the system is both user-friendly and accurate. This system has proven effective in bridging the language barrier and analytical complexity, thereby accelerating literature review processes and improving access to scientific knowledge for the academic community in Indonesia.

    Baca Full Text klik disini
    [thumbnail of S_MKB_2101216_Title.pdf] Text
    S_MKB_2101216_Title.pdf
    Download (677kB)
    [thumbnail of S_MKB_2101216_Chapter 1.pdf] Text
    S_MKB_2101216_Chapter 1.pdf
    Download (285kB)
    [thumbnail of S_MKB_2101216_Chapter 2.pdf] Text
    S_MKB_2101216_Chapter 2.pdf
    Restricted to Staf Perpustakaan
    Download (793kB)
    [thumbnail of S_MKB_2101216_Chapter 3.pdf] Text
    S_MKB_2101216_Chapter 3.pdf
    Download (1MB)
    [thumbnail of S_MKB_2101216_Chapter 4.pdf] Text
    S_MKB_2101216_Chapter 4.pdf
    Restricted to Staf Perpustakaan
    Download (2MB)
    [thumbnail of S_MKB_2101216_Chapter 5.pdf] Text
    S_MKB_2101216_Chapter 5.pdf
    Download (203kB)
    [thumbnail of S_MKB_2101216_Appendix.pdf] Text
    S_MKB_2101216_Appendix.pdf
    Restricted to Staf Perpustakaan
    Download (390kB)
    Official URL: https://repository.upi.edu/
    Item Type: Thesis (S1)
    Additional Information: https://scholar.google.com/citations?user=9v2FBRQAAAAJ&hl=en&oi=ao ID Sinta Dosen Pembimbing: Liptia Venica: 6779029 Mahmudah Salwa Gianti: 6779018
    Uncontrolled Keywords: Asisten Virtual, Large Language Model (LLM), LangChain, Analisis Komparatif, Jurnal Ilmiah, Natural Language Processing (NLP), GPT-4o mini. Virtual Assistant, Large Language Model (LLM), LangChain, Comparative Analysis, Scientific Journal, Natural Language Processing (NLP), GPT-4o mini.
    Subjects: A General Works > AI Indexes (General)
    Q Science > QA Mathematics > QA76 Computer software
    T Technology > T Technology (General)
    T Technology > T Technology (General) > T201 Patents. Trademarks
    Divisions: UPI Kampus Purwakarta > S1 Mekatronika dan Kecerdasan Buatan
    Depositing User: Nurhikam
    Date Deposited: 28 Jul 2025 06:50
    Last Modified: 28 Jul 2025 06:50
    URI: http://repository.upi.edu/id/eprint/134716

    Actions (login required)

    View Item View Item