Taxonomía de algoritmos basados en Machine Learning aplicados en la Ingeniería de Software

Abstract

The constant growth of the software industry has driven companies to explore new ways to improve their processes, generating novel techniques to optimize the tasks involved in software development, in order to increase the efficiency of these processes. At the same time, the terms “Artificial Intelligence” and “Machine Learning” (ML), are being increasingly used, but there still is a certain lack of knowledge about these concepts. Given this context, our main objective is to establish a connection between these disciplines, in order to better understand the benefit of using ML in Software Engineer. In this work, a systematic analysis of the scientific literature published between 2018 and 2023 has been carried out in order to create a taxonomy of Machine Learning algorithms applied to the stages required for software development. The most prominent results indicate that the testing phase in the software development cycle is one of the most researched areas in relation to the aforementioned challenges. Furthermore, it has been observed that some ML algorithms such as Random Forest demonstrate acceptable performance in optimizing one or more tasks simultaneously in the software development process.