Wednesday, November 9th, 2022, 10:00
While Predictive Maintenance approaches are becoming more commonplace as Industry 4.0 makes digitalization of equipment accessible, they still rely on the constant acquisition of large amounts of data, preferably from heterogeneous sources. While ontologies have been utilized in these scenarios for the semantization of data, interoperability and reasoning, the time-sensitive nature of the data is often left out of devised solutions, and the ontologies created to describe the domains remain static. In a scenario in which it is quite common for data scientists to run different analyses with data, often incurring into feature engineering processes, it is very possible for the domain description provided by the ontologies to become outdated. Furthermore, the insights provided by these analyses may also result in new knowledge that could be incorporated into the domain, as to help with future similar endeavors. As such, I propose TICO — the TIme Constrained instance-guided Ontology evolution tool — a tool/framework for ontology evolution that analyses streams or batches of instances as they are generated, and attempts to identify potential changes to their definitions. To ensure that each ontology instance is classified according to the version of the class that makes sense for its time-frame, I also advance the idea of TimeSlices of classes: a copy of how a class "looks like" during a given period, with the restrictions that are valid during that time period.