DBIS

The concept of the Asset Administration Shell (AAS) was developed in the context of “Industrie 4.0” and Smart Manufacturing. It represents a standardized and centralized way of accessing all information on an object (‘asset’; e.g., a machine or a product) in a company. In the AAS, individual properties of the represented asset can be annotated and semantically described using semantic identifiers. These identifiers are based, for example, on dictionaries such as ECLASS and CDD, but in some cases also refer to OWL ontologies.
One of the problems that arise and hinder interoperability is the automated recognition of semantic equivalence of these properties. In preliminary work, a Semantic Matching Service has already been developed that receives matching information from OWL and NLP sentence embedding semantic matchers that have also been developed.
A major challenge with digital twins is maintaining data sovereignty over potentially sensitive company data. A centralized collection of matching information is not realistic. For this reason, in practice, there will be several instances of the Semantic Matching Service that contain different matching information and may source their matching information from different semantic matching technologies. Several instances of this service together represent a distributed graph of matching information.
The goal of this thesis is to investigate methods for visualizing and preprocessing such distributed graphs to empower users in selecting the optimal match. To this end, various approaches—such as graph visualization techniques and interaction with large language models—shall be explored. The initial phase will involve a literature review on semantic matching, graph visualization, and data preprocessing. On that basis, a requirement analysis for a Semantic Matching Visualization Tool is to be carried out, before a proof-of-concept of such a tool is implemented.