Building a banking system specification using machine learning

Abstract

Transforming user requirements into software specification is a complex and demanding task. Artificial intelligence methods such as machine learning (ML) can assist in the software specification process by providing support to system designers. This paper presents an approach based on explanation-based learning (EBL), a ML technique in which a concept is learned by building an explanation. The approach is presented in the context of the system LISE (Learning in Software Engineering). LISE converts a user requirement for a software module into an operational module definition using EBL with an incomplete theory. An example where LISE is used to build the specification of a banking system is illustrated.<>