Abstract: Data collected in Industry 4.0 applications must be converted into tabular datasets before they can be processed by analysis algorithms, as in any data analysis system. To perform this transformation, data scientists have to write complex and long scripts, which can be a cumbersome process. To overcome this limitation, a language called Lavoisier was recently created to facilitate the creation of datasets. This language provides high-level primitives to select data from an object-oriented data model describing data available in a context. However, industrial engineers might not be used to deal with this kind of model. So, this work introduces a new set of languages that adapt Lavoisier to work with fishbone diagrams, which might be more suitable in industrial settings. These new languages keep the benefits of Lavoisier, reducing dataset creation complexity by 40% and up to 80%, and outperforming Lavoisier in some cases.

Authorship

Download reference