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

As the applications of machine learning algorithms in various fields are widely demanded, the development of machine learning software systems (MLS) is rapidly increasing. The quality of MLS is different from that of conventional software systems, in the sense that it depends on the amount and distribution of training data in a model learning and input data during operation. This is a major challenge in quality assurance of MLS development for the enterprise. In this paper, we propose a requirements-driven method to determine the quality characteristics of the MLS. Major contributions of this paper include: (1) Extending the quality characteristics of ISO 25010, which defines the conventional software quality, to those unique to MLS; this paper also defines its measuring method. (2) A method to identify requirements, i.e., issues to be determined in the requirements definition, in order to derive the quality characteristics and measurement methods for MLS, since the quality characteristics and the measurement method depend on the goals of the system under development. In order to evaluate the proposed method, we carried out an empirical study of the quality characteristics and measurement methods related to functional correctness and the maturity of the MLS for the enterprise. Based on the study, we compare the quality characteristics and measurement methods derived by the proposed method with those suggested by developers, and demonstrate the effectiveness of the proposed method.

Abstract Eulerian Method of Moment (MoM) solvers are gaining popularity for multi‐phase CFD simulation involving bubbles or droplets in process engineering. Because the actual positions of bubbles are uncertain, the spatial distribution of bubbles is described by scalar fields of moments, which can be interpreted as probability density functions. Visualizing these simulation results and comparing them to physical experiments is challenging, because neither the shape nor the distribution of bubbles described by the moments lend themselves to visual interpretation. In this work, we describe a visualization approach that provides explicit instances of the bubble distribution and produces bubble geometry based on local flow properties. To facilitate animation, the instancing of the bubble distribution provides coherence over time by advancing bubbles between time steps and updating the distribution. Our approach provides an intuitive visualization and enables direct visual comparison of simulation results to physical experiments.