Martin Ringsquandl

Capturing Industrial Information Models with Ontologies and Constraints

This paper describes the outcomes of an ongoing collaboration between Siemens and the University of Oxford, with the goal of facilitating the design of ontologies and their deployment in applications. Ontologies are often used in industry to capture the conceptual information models underpinning applications. We start by describing the role that such models play in two use cases in the manufacturing and energy production sectors. Then, we discuss the formalisation of information models using ontologies, and the relevant reasoning services. Finally, we present SOMM—a tool that supports engineers with little background on semantic technologies in the creation of ontology-based models and in populating them with data. SOMM implements a fragment of OWL 2 RL extended with a form of integrity constraints for data validation, and it comes with support for schema and data reasoning, as well as for model integration. Our preliminary evaluation demonstrates the adequacy of SOMM’s functionality and performance.

Diagnostics of Trains with Semantic Diagnostics Rules

Industry today employs rule-based diagnostic systems to minimize the maintenance cost and downtime of equipment. Rules are typically used to process signals from sensors installed in equipment by filtering, aggregating, and combining sequences of time-stamped measurements recorded by the sensors. Such rules are often data-dependent in the sense that they rely on specific characteristics of individual sensors and equipment. This dependence poses significant challenges in rule authoring, reuse, and maintenance by engineers especially when the rules require domain knowledge. In this work we propose an approach to address these problems by relying on the well-known Ontology-Based Data Access approach: we propose to use ontologies to mediate the sensor signals and the rules. To this end, we propose a semantic rule language, SDRL, where signals are first class citizens. Our language offers a balance of expressive power, usability, and efficiency: it captures most of Siemens data-driven diagnostic rules, significantly simplifies authoring of diagnostic tasks, and allows to efficiently rewrite semantic rules from ontologies to data and execute over data. We implemented our approach in a semantic diagnostic system and evaluated it. For evaluation we developed a use case of rail systems at Siemens and conducted experiments to demonstrate both usability and efficiency of our solution.

Event-Enhanced Learning for KG Completion

Statistical learning of relations between entities is a popular approach to address the problem of missing data in Knowledge Graphs. In this work we study how relational learning can be enhanced with background of a special kind: event logs, that are sequences of entities that may occur in the graph. Events naturally appear in many important applications as background. We propose various embedding models that combine entities of a Knowledge Graph and event logs. Our evaluation shows that our approach outperforms state-of-the-art baselines on real-world manufacturing and road traffic Knowledge Graphs, as well as in a controlled scenario that mimics manufacturing processes.

Knowledge Graph Constraints for Multi-label Graph Classification

Graph classification methods have gained increasing attention in different domains, such as classifying functions of molecules or detection of bugs in software programs. Similarly, predicting events in manufacturing operations data can be compactly modeled as graph classification problem. Feature representations of graphs are usually found by mining discriminative sub-graph patterns that are non-uniformly distributed across class labels. However, as these feature selection approaches are computationally expensive for multiple labels, prior knowledge about label correlations should be exploited as much as possible. In this work, we introduce a new approach for mining discriminative sub-graph patterns with constraints that are extracted from links between labels in knowledge graphs which indicate label correlations. The incorporation of these constraints allows to prune the search space and ensures extraction of consistent patterns. Therefore, constraint checking remains efficient and more robust classification results can be obtained. We evaluate our approach on both, one public and one custom simulated data set. Evaluation confirms that incorporation of constraints still results in efficient pattern mining and can increase performance of state-of-the-art approaches.