Abstract

In this contribution, we give an insight in our experiences in the technical and organizational realization of industrial analytics. We address challenges in implementing industrial analytics in real-world applications and discuss aspects to consider when designing a machine learning solution for production. We focus on technical and organizational aspects to make industrial analytics work for real-world applications in factory automation. As an example, we consider a machine learning use case in the area of industry compressors. We discuss the importance of scalability and reusability of data analytics pipelines and present a container-based system architecture. Document type: Part of book or chapter of book

Abstract

An updated metric developed to assess the degree of open data reusability, called MEtric for the evaLuation of Open DAta: Meloda 5 is presented. Previous version of the metric, Meloda 4, had six dimensions: the legal licensing of data, the mechanisms to access the data, the technical standards of the datasets, the data model, the geographic content of the data and the updating frequency. With all these dimensions, the metric provides a quantitative evaluation about how reusable the datasets released are. During the last five years, this metric has been cited and used by some other authors but the extensive use of the metric has brought to light some of its limitations. In order to get deeper insights into the topic, a panel of international experts has been surveyed about two aspects of the metric. First aspect was what other factors should be considered in order to qualify the reusability of a released dataset. And the second aspect was the internal structure, the levels for every dimension of the metric; if they should be increased, merged, removed or divided. Considering the results of the survey, first, we identify the factors / dimensions that should be kept: legal licensing, access to information, technical standard, standardization, geolocation content and updating frequency of data. Second, we consider the inclusion of two new dimensions: dissemination and reputation. Then, we present the new internal structure, the levels for each dimension, and the measures to evaluate the degree of reuse of each dataset. Finally, a standardization of the metric for other steps of the data impact process, data reuse analytics and data-driven services generation are presented together with future research lines.