PETIR

Software Quality in The Use-based Context: A Systematic Literature Review

Mochamad Farid Rifai — 2026-10-14

One crucial aspect to consider when measuring software quality is the use-based context. This approach allows for results that align with user needs. Measuring software quality within this context combines selected characteristics from Software Quality-in-Use (QiU) and Software Product Quality (PQ). However, there remains a limited body of literature on measuring software quality in this way. This study seeks to perform a Systematic Literature Review (SLR) to identify research trends, employed methodologies, and key findings from existing literature. The preparation and execution of this SLR adhered to the PRISMA guidelines. Our analysis included papers published between 2014 and 2024 from four digital library databases (Scopus, Science Direct, Springer, and Wiley), totaling 738 papers. After a thorough screening and eligibility assessment, we included 36 studies in our review. Our comprehensive analysis of previously conducted research revealed key findings related to the integration of software quality with Machine Learning (ML), Decision Support Systems (DSS), Goal-Question-Metric (GQM), and various implementation case studies. This focus offers a broader perspective on measuring software quality in the use-based context.

DYNAMIC VM AUTOSCALING IN CLOUD COMPUTING USING HONEY BADGER ALGORITHM

Muhammad Hasan Basri — 2026-10-14

Komputasi awan memaksa manajemen sumber daya yang efisien untuk menangani fluktuasi beban kerja yang dinamis. Metode autoscaling berbasis ambang batas sering mengalami keterlambatan respons penyediaan yang menyebabkan pelanggaran service level aggreement (SLA), terutama saat terjadi lonjakan trafik secara tiba-tiba. Penelitian ini mengusulkan autoscaling proaktif menggunakan honey badger algorithm untuk mengoptimalkan alokasi container secara dinamis. Fungsi fitness multi objektif pada HBA dilakukan modifikasiuntuk mengintegrasikan penalti SLA. Efisiensi sumber daya dan biaya operasional untuk mendeteksi risiko pelanggaran yang lebih awal. Validasi dilakukan melalui pendekatan emulasi berbasis docker pada lingkungan wsl dengan membandingkan kinerja sistem pada skenario beban normal dan beban ekstrem. Hasil eksperimen menunjukkan bahwa pada beban normal, HBA mampu mempertahankan rata-rata waktu respons sebesar 0,29 detik dengan prioritas performa. Pada skenario ekstrem algoritma menunjukkan resiliensi tinggi dengan membatasi tingkat pelanggaran SLA sebesar 20% dan menjaga utilitas CPU pada 49,10% menggunakan rata-rata 1,98 kontainer. Penelitian ini membuktin bahwa HBA mampu menyeimbangkan trade-off antara kualitas layanan dan efisiensi biaya secara otonom pada lingkup sumber daya yang terbatas.

Text-to-Speech Bahasa Melayu Dialek Lubuk Kandis Menggunakan Metode VITS

Dedi Erlyanto — 2026-10-14

The existence of numerous regional languages in Indonesia reflects the country's cultural richness, with their historical distribution forming various dialects. One of them is the Malay Language of the Lubuk Kandis Dialect. To support the preservation and dissemination of this language, this study applies speech synthesis technology in the form of Text-to-Speech (TTS) to generate speech output for the Lubuk Kandis dialect. A dataset consisting of 500 daily activity sentences was recorded by a native speaker of Lubuk Kandis and validated by a local linguistic expert. The Variational Inference with Adversarial Learning for End-to-End Text-to-Speech (VITS) method was adopted to produce speech output that resembles natural human voice. The evaluation used the Mean Opinion Score (MOS) test involving five respondents and resulted in an average score of 4.72, indicating that the generated speech falls into the good category. These findings demonstrate that the VITS method is effective in synthesizing natural and intelligible speech for the Malay Language of the Lubuk Kandis Dialect, providing potential support for the preservation of regional languages.

Sistem Prediksi Kebutuhan Bandwidth dengan Machine Learning Metode Long Short Term Memory (LSTM)

Ananta Wicaksana — 2026-10-14

Internet has become an essential need in various aspects of life, especially in public environments such as Malang Creative Centre (MCC). However, bandwidth management that is still done manually at MCC causes inefficient allocation and has an impact on service quality. Therefore, this research aims to design a bandwidth usage prediction system using the Long Short Term Memory (LSTM) method to optimise bandwidth management at MCC. The LSTM method was chosen because of its ability to learn past data patterns and provide accurate predictions. This research uses 80% training data and 20% test data, with the optimal LSTM model configuration consisting of 80 layers and 100 epochs, and optimisation using Adam. The evaluation results show a prediction accuracy rate of 82.39%, which indicates that this model is reliable in predicting bandwidth usage on various floors and times that have been determined. In conclusion, this LSTM-based prediction system offers a solution for more precise and efficient bandwidth management at MCC. This not only saves network operational costs but also improves the quality of internet services for users, supporting creative economic activities at MCC by providing adequate and efficient internet connections.

Metode Klasifikasi Support Vector Machine (SVM) Untuk Analisis Sentimen Aplikasi Bing: Chat with AI & GPT-4 Di Google Play Store

Muslih — 2024-06-07

The development of Artificial Intelligence (AI), particularly through ChatGPT technology and the emergence of Bing: Chat with AI & GPT-4, forms the foundation for this research. This study conducts sentiment analysis on users of the Bing: Chat with AI & GPT-4 application using the Support Vector Machine (SVM) classification method. A total of 10,351 reviews from the Google Play Store underwent preprocessing stages, including cleaning, case-folding, tokenizing, stop word removal, stemming, and filtering. The implementation of the TF-IDF weighting method and Support Vector Machine (SVM) classification resulted in a negative sentiment of 58.37% and a positive sentiment of 41.66%, with a testing accuracy of 89.44%. Visualization using word clouds, incorporating common words in positive and negative reviews, provides insights into user preferences and dissatisfaction, particularly regarding the speed and performance of Bing: Chat with AI & GPT-4. The study concludes that improving the application's quality is crucial for future user satisfaction, and future research recommendations include utilizing data from other platforms such as the App Store and social media for comprehensive sentiment analysis.