OntoTrans – Review of the Project’s First Year

OntoTrans is a European research project funded by Horizon 2020, which responds to the need of industry to approach manufacturing challenges more efficiently. The effective reuse of manufacturing knowledge and the effective utilisation of materials modelling are crucial here. Thus, end users ought to be supported throughout the translation process of a manufacturing challenge. OntoTrans aims to achieve this by providing a general-purpose and ontology-based Open Translation Environment (OTE) on-top of which apps for smart guidance can be build. Three companies from industry are involved in the project by providing innovation cases for manufacturing, namely P&G, Composite Evolution, and ArcelorMittal.

The project started last year, and has already a number of results to account for. A core component of the translation environment are formal ontologies that cover the domain of interests of the industrial applications. A team around Emanuele Ghedini from the University of Bologna is building these ontologies based on the European Materials & Modelling Ontology (EMMO). They are designed collaboratively with domain experts from the partnering companies in weekly hackathons using Miro boards. Eventually, these application ontologies are supposed to be integrated into a knowledge base with the corresponding manufacturing metadata. Stardog has been chosen as solution for the knowledge base, due to the fact that it supports a wide range of reasoning profiles, has a sophisticated user interface, and has already been applied successfully in other projects of partners in OntoTrans. Additionally, an architecture document has been released as first draft for the OTE environment, which has three main modules, that is, (1) a knowledge base, (2) application and tools for smart guidance and (3) data and simulation services. A major concern is the access mechanism to the different services in the OTE as well as the industrial data by all the different applications in the OTE environment. This is solved with a (functional) pipeline pattern in which the accessed information is translated into an intermediate format named Dlite, mapped, filtered, and transformed.

We, at the SemSys group at TU Wien, are especially involved in providing an exploratory search interface over the resulting knowledge graph. This search interface shall assist end user of the industrial apps in finding and configuring models (e.g. for machine learning or simulations). Moreover, it shall support the end user in analysing and making sense of modelling results. Most importantly, we designed mockups of how the interaction between industrial apps and our exploratory search prototype could look like.

At the end of September 2021, a review meeting of the research project was conducted, and the two present reviewers were providing valuable feedback being in highly satisfied with the current project progress.