Software Engineering for Data Mining (ML-Enabled) Software Applications

Abstract

As the data increase keeps on getting more extensive due to technology evolvement from the rational database, online transaction, cloud computing, data warehouse to big data analytics. This changes influences organizations to advance from data mining support to machine learning-enabled software platform. Seemingly, the study summarised secondary data from non-grey and grey academic literature as the research field recently started getting attention. Consequently, the work identifies, analyzes, and synthesizes the challenges of ML-enabled software development, which differs from traditional software development. But, with the adoption of the SE technique to engineer ML-enabled software development, the study was able to identify advancement for ML-enabled software likes automation of mismatch detection, which occurs due to the nature of different perspectives of stakeholders involved. Another one is integrating ML and SE data end-to-end pipeline to allow Systematic test mechanism and test automation where necessary when ML is complex in format to enable standard SE test logs. Then, education, training, and cooperation between the stakeholders, especially SE and ML, to gain more experience, knowledge, put rifts aside to join hands, and work together to ascertain user requirements. Finally, the work reframed the traditional SE development process to engineer the ML software development process. Therefore, the study can benefit stakeholders in the ML and SE communities in handling ML development challenges and may benefits academicians in conduction future research on software engineering for artificial intelligence.