Transforming software size estimation with prompt engineering: A ChatGPT-based framework for component-based systems

Abstract

In software project management, it is essential to estimate software size accurately since knowing the size of the program allows for more efficient p lanning, e stimation, and scheduling of its development. The conventional software size estimation approaches such as use of Function Points (FP) and its extensions often prove to be time consuming, resource-intensive, and thereby a costly exercise, demanding specialized human expertise. This has no difference when it comes to the modern software development paradigms like Component-Based Systems Development (CBSD). On the other hand, concerning Artificial Intelligence (AI), the large language model-based chat-bots like ChatGPT, Bard AI, DALL-E, and Midjourney are excelling at automating traditional human activities and interactions in various domains including software engineering. Among all these AI tools, ChatGPT has proven its applicability in many industries including software engineering, acquiring around 100% accuracy over other chat-bots. In this paper, we therefore developed and validated an innovative framework based on AI, to measure the size of a Component-Based Systems (CBS) using ChatGPT. The framework which consists of a set of prompts has been designed to expedite the size estimation process of a CBS using an extension of FP called Component Point (CP) while substantially reducing the need of human involvement and financial o utlay. O ur a im i s n ot o nly t o e nhance t he efficiency of software size estimation but also to conserve both time and financial r esources t hat w ould o therwise b e s pent o n practicing conventional approaches. We therefore envision that the proposed approach woul