Construction of a quality model for machine learning systems

Abstract Nowadays, systems containing components based on machine learning (ML) methods are becoming more widespread. In order to ensure the intended behavior of a software system, there standards that define necessary qualities system and its (such as ISO/IEC 25010). Due different nature ML, we have re-interpret existing for ML or add new ones trustworthiness). We be very precise about which quality property is relevant entity interest completeness training data correctness trained model), how objectively evaluate adherence requirements. this article, present systematically construct models an industrial use case. This model enables practitioners specify assess objectively. addition overall construction process described, main outcomes include meta-model specifying systems, reference elements regarding views, entities, properties, measures research, example instantiation concrete case, lessons learned from applying process. found it crucial follow systematic in come up with measurable properties can evaluated practice. future, want learn term differs between types evaluating systems.