Towards A Visualisation Ontology for Reusable Visual Analytics

Abstract

Data analytics including machine learning analytics is essential to extract insights from production data in modern industries. Visual analytics is essential for data analytics for e.g., presenting the data to provide an instinctive perception in exploratory data analysis, facilitating the presentation of data analysis results and the subsequent discussion on that. Visual analytics should allow a transparent common ground for discussion between experts in data analysis projects, given the multidisciplinary background of these experts. However, a standarised and formalised way of describing the knowledge and practice of visualisation is still lacking in the industry, which hamstrings the transparency and reusability of visual analytics. A visualisation ontology which models the nature and procedure of visualisation is well-suited to provide such standardisation. Currently a few studies discuss partially the modelling of visualisation, but insufficiently study the procedure of visualisation tasks, which is important for transparency and reusability especially in an industrial scenario. To this end, we present our ongoing work of development of the visualisation ontology in industrial scenarios at Bosch. We also demonstrate its benefits with case studies and knowledge graph based on our ontology.