Functional Size Measurement With Conceptual Models: A Systematic Literature Review

ABSTRACT The demand for efficient functional size measurement (FSM) methods in the competitive software market today is undeniable. However, incomplete and imprecise system specifications pose significant challenges, particularly in scenarios that require fast, flexible, and accurate software size estimation, such as public tenders. Although the integration of conceptual models within FSMs offers a promising solution to these issues, a systematic exploration of such methods remains largely unexplored. This work evaluates FSM methods that integrate conceptual models by analyzing studies from the past 20 years. It highlights key contributions and advances in proposed conceptual model‐based FSM methods. In addition, the study examines their limitations and challenges, offering insights for future improvements. A systematic literature review (SLR) was conducted to guide the research process. The review was organized around three research questions, each targeting the study's key objectives: (1) to explore FSM methods utilizing conceptual models, (2) to summarize proposals for their improvement, and (3) to identify the limitations of the proposed enhancements. Primary studies span two decades (2004–2024), with peaks in 2008 and 2015, averaging one to two studies annually. Of the 1371 initial studies, 13 were selected using strict criteria. These studies are categorized into Measurement Techniques (30.77%), Automation (38.46%), and Application‐Specific topics (30.77%). The contributions of the primary studies are analyzed in terms of their approaches Repeatability and Validation . Repeatability is assessed by examining whether the primary studies proposed a formal model when using real datasets. In contrast, Validation focuses on whether the studies were tested in real‐world projects. A total of 46.15% of the primary studies utilize formal models, whereas 53.85% rely on nonformal models, although dataset size is often unspecified. Most studies validate their methods using 1 to 30 projects. Common Software Measurement International Consortium (COSMIC) is the most widely used FSM method (69.23%), followed by the Function Point Analysis (FPA) (15.38%) and custom Methods (15.38%), with conceptual UML models appearing in 84.61% of the studies. Key limitations, including Scalability and Generalizability , Complexity Robustness , and Flexibility , persist across all categories. Notably, Scalability and Generalizability was identified as a limitation in 75% of Measurement Techniques studies, 80% of Automation studies, and 75% of Application‐Specific studies, while Flexibility challenges were most pronounced, affecting 100% of Application‐Specific studies. The limited number of primary studies underscores a substantial research gap in conceptual model‐based FSM methods. Future research should focus on developing formal models to enhance theoretical rigor, leveraging real‐world datasets for validation, providing comprehensive methodological descriptions, and standardizing validation practices. Additionally, prioritizing advancements in FSM methods by improving scalability , generalizability , and flexibility is crucial. These enhancements will enable FSM methods to effectively manage complex systems, adapt across diverse software domains, and address application‐specific requirements, ensuring their continued relevance in dynamic and evolving software development environments.