Claim Missing Document
Check
Articles

Found 2 Documents
Search

Studi Literatur Penerapan Topic Modeling Menggunakan Latent Dirichlet Allocation (LDA) dalam Analisis Dokumen dan Laporan Publik Citra Grace Kalangie; Irene R. H. T. Tangkawarow
Edutik : Jurnal Pendidikan Teknologi Informasi dan Komunikasi Vol. 6 No. 1 (2026): EduTIK : Februari 2026
Publisher : Jurusan PTIK Universitas Negeri Manado

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Perkembangan teknologi informasi menyebabkan peningkatan signifikan jumlah dokumen teks tidak terstruktur, khususnya dalam bentuk laporan dan dokumen publik. Kondisi ini menimbulkan tantangan dalam proses analisis manual yang tidak lagi efisien dari sisi waktu dan sumber daya. Salah satu pendekatan yang banyak digunakan untuk mengatasi permasalahan tersebut adalah topic modeling berbasis text mining, dengan Latent Dirichlet Allocation (LDA) sebagai metode yang paling populer. Penelitian ini bertujuan untuk mengkaji secara sistematis penerapan metode LDA dalam analisis dokumen dan laporan publik berdasarkan studi-studi terdahulu. Metode penelitian yang digunakan adalah studi literatur dengan menelaah artikel ilmiah dari jurnal nasional dan internasional yang relevan. Hasil kajian menunjukkan bahwa LDA efektif dalam mengidentifikasi topik laten pada data teks tidak terstruktur di berbagai konteks, termasuk media sosial, dokumen akademik, dan laporan institusional. Namun demikian, keterbatasan masih ditemukan pada aspek interpretasi topik dan penentuan jumlah topik yang optimal. Studi ini diharapkan dapat menjadi rujukan konseptual bagi penelitian lanjutan, khususnya dalam pemanfaatan LDA untuk analisis dokumen resmi pemerintahan.  
Implementation of Latent Dirichlet Allocation for the Analysis of Election Logistics Distribution Reports at Bawaslu RI Citra Grace Kalangie; Irene Realita Tangkawarouw; Audy Kenap
Journal of Embedded Systems, Security and Intelligent Systems Vol 7 No 4 (2026): December 2026
Publisher : Program Studi Teknik Komputer

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.59562/jessi.v7i4.12849

Abstract

Purpose – This study aims to implement Latent Dirichlet Allocation (LDA) for analyzing election logistics distribution monitoring reports at Bawaslu RI and to develop an interactive dashboard to support topic exploration. The study addresses the difficulty of manually interpreting large collections of institutional reports containing operational, administrative, geographical, procurement, and legal information related to election logistics distribution. Design/methods/approach – This research applied a quantitative text mining approach. A total of 254 valid PDF-based monitoring reports were extracted, validated, cleaned, and transformed into an analyzable corpus. Text preprocessing involved case folding, text cleaning, tokenization, and stopword removal using Indonesian and domain-specific stopword lists. The cleaned corpus was represented using Bag of Words and modeled using LDA with the Gensim library. Topic configurations from 2 to 15 were evaluated using the C_V coherence score, and the results were integrated into a web-based dashboard using Streamlit and PyLDAvis. Findings – The optimal model produced ten topics with a C_V coherence score of 0.4040. The generated topics represented themes such as regional supervision, logistics completeness, administrative procedures, island distribution, procurement, archipelago distribution, and post-election legal processes. The dashboard enabled interactive exploration of topics, coherence evaluation, topic interpretation, and visualization of intertopic relationships. Research implications/limitations – The study demonstrates the value of topic modeling for improving the analysis of government monitoring reports, although expert validation with Bawaslu officers remains necessary. Originality/value – This study contributes by applying LDA to official election logistics reports and integrating the results into an accessible analytical dashboard for election supervision stakeholders.