Industrial engineering is an interdisciplinary activity that involves human experts from various technical backgrounds working with different engineering tools. In the era of digitization, the engineering process generates a vast amount of data. To store and exchange such data, dedicated international standards are developed, including the XML-based data format AutomationML (AML). While AML provides a harmonized syntax among engineering tools, the semantics of engineering data remains highly heterogeneous. More specifically, the AML models of the same domain or entity can vary dramatically among different tools that give rise to the so-called semantic interoperability problem. In practice, manual implementation is often required for the correct data interpretation, which is usually limited in reusability.
Efforts have been made for tackling the semantic interoperability problem. One mainstream research direction has been focused on the semantic lifting of engineering data using Semantic Web technologies. However, current results in this field lack the study of building complex domain knowledge that requires a profound understanding of the domain and sufficient skills in ontology building. ... mehrThis thesis contributes to this research field in two aspects. First, machine learning algorithms are developed for deriving complex ontological concepts from engineering data. The induced concepts encode the relations between primitive ones and bridge the semantic gap between engineering tools. Second, to involve domain experts more tightly into the process of ontology building, this thesis proposes the AML concept model (ACM) for representing ontological concepts in a native AML syntax, i.e., providing an AML-frontend for the formal ontological semantics. ACM supports the bidirectional information flow between the user and the learner, based on which the interactive machine learning framework AMLLEARNER is developed.
Another rapidly growing research field devotes to develop methods and systems for facilitating data access and exchange based on database theories and techniques. In particular, the so-called Query By Example (QBE) allows the user to construct queries using data examples. This thesis adopts the idea of QBE in AML-based engineering by introducing the AML Query Template (AQT). The design of AQT has been focused on a native AML syntax, which allows constructing queries with conventional AML tools. This thesis studies the theoretical foundation of AQT and presents algorithms for the automated generation of query programs. Comprehensive requirement analysis shows that the proposed approach can solve the problem of semantic interoperability in AutomationML-based engineering to a great extent.