<![CDATA[Transformasi digital mendorong organisasi pelayanan publik untuk menyediakan layanan yang semakin mudah, cepat, andal, dan responsif terhadap kebutuhan pengguna. Penelitian ini bertujuan menganalisis distribusi sentimen pengguna PLN Mobile, mengidentifikasi topik utama melalui Latent Dirichlet Allocation (LDA), serta memetakan temuan ke dalam dimensi E-Service Quality. Penelitian menggunakan desain kuantitatif deskriptif dengan pendekatan text mining terhadap 500 ulasan bersih pengguna PLN Mobile pada Google Play Store tahun 2026 yang dikumpulkan melalui web scraping. Data diproses melalui cleaning, case folding, tokenizing, stopword removal, dan stemming menggunakan Python dan Sastrawi. Klasifikasi sentimen dilakukan berdasarkan polaritas rating dan diverifikasi menggunakan leksikon bahasa Indonesia, sedangkan LDA digunakan untuk mengidentifikasi struktur topik laten. Hasil penelitian menunjukkan bahwa 99,8% ulasan termasuk sentimen positif, terutama berkaitan dengan kemudahan pembelian token dan kecepatan pembayaran tagihan. Pemodelan topik menunjukkan dominasi aspek efficiency, responsiveness, fulfillment, dan system availability. Meskipun sentimen negatif hanya 0,2%, ulasan tersebut mengindikasikan kendala autentikasi login dan stabilitas sistem. Temuan ini menunjukkan bahwa integrasi analisis sentimen, LDA, dan E-Service Quality mampu menghasilkan evaluasi layanan digital yang lebih komprehensif serta mendukung pengambilan keputusan berbasis data. Namun, hasil perlu ditafsirkan dengan mempertimbangkan potensi self-selection bias pada ulasan pengguna. Pendekatan ini juga membantu mengidentifikasi prioritas perbaikan layanan secara lebih objektif dan berkelanjutan. Sentiment Analysis and Topic Modeling Of Pln Mobile App Reviews: Integrating LDA Method and E-Service Quality Dimensions Abstract Digital transformation has encouraged public service organizations to provide services that are increasingly accessible, fast, reliable, and responsive to user needs. This study aims to analyze the distribution of user sentiment toward PLN Mobile, identify major topics using Latent Dirichlet Allocation (LDA), and map the findings onto E-Service Quality dimensions. The study employed a descriptive quantitative design with a text mining approach using 500 cleaned user reviews of PLN Mobile collected from the Google Play Store in 2026 through web scraping. The data were processed through cleaning, case folding, tokenizing, stopword removal, and stemming using Python and Sastrawi. Sentiment classification was conducted based on rating polarity and cross-validated using an Indonesian sentiment lexicon, while LDA was applied to identify latent topic structures. The results showed that 99.8% of the reviews were classified as positive sentiment, mainly associated with ease of token purchase and speed of bill payment transactions. Topic modeling revealed the prominence of efficiency, responsiveness, fulfillment, and system availability dimensions. Although negative sentiment accounted for only 0.2%, these reviews indicated issues related to login authentication and system stability. These findings demonstrate that integrating sentiment analysis, LDA, and E-Service Quality can provide a more comprehensive evaluation of digital services and support data-driven decision-making. However, the findings should be interpreted cautiously due to potential self-selection bias in user reviews. This approach also helps identify service improvement priorities more objectively and sustainably.]]>
