Berikut beberapa paper yang sudah berhasil dipublikasikan pada Juli 2025:
Exploring Innovative Approaches for Software Development Risk Assessment and Management
Setya Uswatun Hasanah, Paulus Insap Santosa, dan Ridi Ferdiana
DOI: 10.1109/ICITISEE63424.2024.10730152
The success of software development projects is significantly impacted by various risk factors, both predictable and unpredictable. This literature review aims to explore and evaluate innovative risk assessment and management techniques in software development. Using the Systematic Literature Review (SLR) method, this study reviews existing research on new software development risk assessment and management methods. This extensive review highlights various approaches, tools, and strategies for identifying, evaluating, and mitigating risks in software projects. The results underscore the importance of creative solutions in tackling risk management challenges. Practitioners are encouraged to adopt innovative methods for risk assessment and management as the field of software development evolves. Furthermore, this study recommends that researchers continue to investigate emerging trends and challenges in risk management. Emphasis is placed on developing hybrid models that combine traditional and innovative approaches, examining risk management practices in new technologies such as AI and machine learning, and addressing cultural and organisational barriers to adopting new techniques. Collaboration and knowledge sharing between practitioners and researchers are expected to improve risk management practices in software development significantly.
House of Risk Phase 1: A Tool for Identifying and Prioritizing Software Development Risk Agents
Setya Uswatun Hasanah, Paulus Insap Santosa, dan Ridi Ferdiana
DOI : 10.1109/ICADEIS65852.2025.10933042
This research explores risk management in software development by applying the House of Risk (HOR) Phase 1 method to pinpoint and rank risk sources. Through interviews with experts in IT, healthcare and government sectors, 30 risk events and 29 risk agents were identified in various phases of the software development life cycle. The HOR model was used to determine each risk agent’s Aggregate Risk Potential (ARP), allowing the prioritization of risks based on their relative impact. Significant findings revealed that planning, task prioritization, and awareness of new software implications are critical areas that need urgent attention. The results demonstrate the effectiveness of the HOR Phase 1 model in optimizing risk management practices, providing practical guidance to project managers and developers to improve project stability and reduce negative impacts. This research advances software risk management by presenting a systematic method for identifying and addressing top-priority risks, thereby fostering more robust software development processes.
Towards Two-Step Fine-Tuned Abstractive Summarization for Low-Resource Language Using Transformer T5
Salhazan Nasution, Ridi Ferdiana, dan Rudy Hartanto
This study explores the potential of two-step fine-tuning for abstractive summarization in a low-resource language, focusing on Indonesian. Leveraging the Transformer-T5 model, the research investigates the impact of transfer learning across two tasks: machine translation and text summarization. Four configurations were evaluated, ranging from zero-shot to two-step fine-tuned models. The evaluation, conducted using the ROUGE metric, shows that the two-step fine-tuned model (T5-MT-SUM) achieved the best performance, with ROUGE-1: 0.7126, ROUGE- 2: 0.6416, and ROUGE-L: 0.6816, outperforming all baselines. These findings demonstrate the effectiveness of task transfer-ability in improving abstractive summarization performance for low-resource languages like Indonesian. This study provides a pathway for advancing natural language processing (NLP) in low-resource language through two-step transfer learning.
DOI: 10.14569/IJACSA.2025.01602120
Large Language Model Employment for Story Point Estimation Problems in AGILE Development
Barkhah Permana, Ridi Ferdiana, dan Azkario Pratama
Software effort estimation (SEE) in Agile Software Development (ASD) has been a longstanding challenge for software engineers. The traditional approach estimates effort scores known as Story Point (SP) to measure user stories. The approach was an expert judgment method in which experts propose certain values of story points based on their experience through various mechanisms such as planning poker, Dot Voting, Bucket System, etc. The process will become challenging if experts have different opinions about the value of certain story points, which may result in a longer duration for completion. Many algorithms have been proposed to optimize the method. The utilization of Artificial Intelligence (AI) and Machine Learning (ML) has greatly simplified processes. Various algorithms are utilized, including Fuzzy, SVM, Naive Bayes, KNN, ANN, RNN, CNN, and LSTM, and utilizing Large Language Models (LLMs) for advancement. This paper employs the hybrid model of Bidirectional Encoder Representations from Transformers (BERT) – Multilayer Perceptron (MLP), which we call alterBERT, to perform regression for estimating Story Points (SP). We will assess the model over the TAWOS dataset and compare it with the previous model, GPT2SP, and other GPT2 architecture. The dataset consists of issues from 39 open-source projects mined from JIRA Software. According to the experimental result, we conclude that the alterBERT models have better performance, outperforming the other models with an average MAE and RMSE, respectively, about 1.72 and 2.51
DOI: 10.1109/ICECOS63900.2024.10791206
Systematic Literature Review of Software Estimation in Global Software Development
Arifia Kasastra, Ridi Ferdiana, dan Syukron Abu Ishaq Alfarozi
Software development is always characterized by certain parameters. In the context of Global Software Development (GSD), one of the key challenges for software developers is predicting the development effort of a software system based on developer details, size, complexity, and other measures. When development teams are spread apart, it presents a number of challenges. Communication and coordination become more complex, leading to hidden costs associated with managing software development across different locations. Hence, the software estimation models used for co-located software development are not suitable for estimation in GSD. Software estimation in GSD has become a crucial area of research. This paper presents the findings of a systematic literature review on software estimation in the context of GSD, with the aim of providing researchers and practitioners with a comprehensive understanding of the current state of the art in this area. Over the last decade, only a handful of researchers have devoted their attention to estimating efforts in GSD. The results show that estimation methods in GSD have not been widely explored. The COCOMO-II method is the most widely adapted method for GSD, and Time Zone is the most agreed Cost Driver for GSD.
DOI: 10.1109/ICITCOM62788.2024.10762359
