A Model of Reusable Assets in AIE Software Systems

Abstract

Nowadays, due to the increasing presence of artificial intelligence in software systems, development teams face the challenge of working together to integrate tasks, resources, and roles in a new field, named AI Engineering. Proposals, in the way of models, high-light the needs of integrating two different perspectives – the software and the decision-making support (big data, machine learning, and so on) systems. But there is something more – both systems must achieve high quality levels for different properties; and this is not a straightforward task. Quality properties, such as reusability, traditionally evaluated and reinforced through modeling in software systems, do not exactly apply similarly in decision-making support systems. In this paper, we propose a model for managing reusable assets in AI engineered systems by linking software product line modeling and variety identification. The proposal is exemplified through a case study in the agriculture domain.