Presentation at ICAICTA 2021 “Using SPARQL to express Causality in Explainable Cyber-Physical Systems”
On Wednesday, 29 September 2021, one paper from the SemSys group was presented at the International Conference on Advanced Informatics: Concepts, Theory and Applications (ICAICTA).
The paper entitled: “Using SPARQL to express Causality in Explainable Cyber-Physical Systems” The authors are: Peb R. Aryan, Matthias Deimel, Fajar J. Ekaputra, Marta Sabou.
The paper is part of the ExpCPS project. In this paper, we presented the various methods to represent causality in the literature and the one that is used in the ExpCPS framework. In the ExpCPS we use a rule to represent causality between two abstract or physical properties. The properties are then observed or actuated using a device placed at particular locations in the actual grid. Explaining the events happening at some location in the grid exploits the topological and causal relationship between the devices.
Even though the representation is quite simple, the amount of work the domain experts need to do to annotate the relationships is considerable as it scales with the size of the grid. Therefore, in this paper we propose an intermediate expression to help automate the work. By augmenting a simple rule and rule constraints expressed as SPARQL query, we have a reusable set of causality rules that can be applied to different grid configurations or it can respond to configuration changes.
This paper is a starting work in the line of causality representation. Future work in this line is to derive or augment the rules with weights such that it can be used to rank explanations.
Also at this conference, Peb was invited to chair one of the sessions and was the first experience being a session chair. The conference was organized as an online event from 8 AM to 6 PM (WIB) for two days (29-30 September). At the end of the conference, it was announced that the next one will be held in Tokoname, Aichi (Japan) and hopefully will be conducted in face-to-face mode.
Keyword: explainability, smart grid, knowledge graph, causality, sparql
