Semantics for Cyber-Physical Systems: A Cross-Domain Perspective

Tracking #: 2275-3488

Marta Sabou
Stefan Biffl
Alfred Einfalt
Lukas Krammer
Wolfgang Kastner
Fajar J. Ekaputra

Responsible editor: 
Guest Editor 10-years SWJ

Submission type: 
Modern life is increasingly made more comfortable, efficient, and sustainable by the smart systems that surround us: smart buildings monitor and adjust temperature levels to achieve occupant comfort while optimizing energy consumption; smart energy grids reconfigure dynamically to make the best use of ad-hoc energy produced by a host of distributed energy producers; smart factories can be reconfigured on the shop-floor to efficiently produce a diverse range of products. These complex systems can only be realized by tightly integrating components in the physical space (sensors, actuators) with advanced software algorithms in the cyber-space, thus creating so-called Cyber-Physical Systems(CPS). Semantic Web technologies (SWT) have seen a natural uptake in several areas based on CPS, given that CPS are data and knowledge intensive while providing advanced functionalities typical of semantics-based intelligent systems. Yet, so far, this uptake has primarily happened within the boundaries of application domains resulting in somewhat disconnected research communities. In this paper, we take a cross-domain perspective by synthesizing our experiences of using SWTs during the engineering and operation of CPS in smart manufacturing, smart buildings and smart grids. We discuss use cases that are amenable to the use of SWTs, benefits and challenges of using these technologies in the CPS lifecycle as well as emerging future trends. While non-exhaustive, our paper aims at opening up a dialog between these fields and at putting the foundation for a research area on semantics in CPS.
Full PDF Version: 

Minor Revision

Solicited Reviews:
Click to Expand/Collapse
Review #1
By Agnieszka Lawrynowicz submitted on 09/Sep/2019
Minor Revision
Review Comment:

The paper presents a cross-domain view on the topic of semantic technologies for Cyber-Physical Systems (CPS).
These complex systems are being built by a tight integration of components in the physical space (e.g., sensors) with advanced software in the cyber-space.
The authors claim that Semantic Web technologies have been mostly applied within particular domains so far, and they take a cross-domain perspective by synthesizing their experiences of using Semantic Web technologies during the engineering and operation of CPS in smart manufacturing, smart buildings and smart grids.
The aspect of heterogeneous data integration is the most stressed in the paper.

The paper is a solid piece of work, providing an overview of CPS in the discussed domains, the use of semantic technologies for CPS, together with examples of applications, use cases, advantages and short-term and long-term research challenges.

Below I provide some remarks related to selected sections:

**** Section 2 ***
This section mostly describes domains and challenges of handling heterogenous, complex data and knowledge, with mostly just mentions on semantic technologies.
Here I lack the definition of what a "digital twin" is - this may be not that obvious for a reader who is not knowledgeable in the manufacturing domain.
It would be also interesting to know more on how Semantic Web technologies are used for digital twins. Maybe even to give an example?

**** Section 3 ***
This section presents use cases in CPS amenable to the use of Semantic Web Technologies.
It also gathers referenced uses of Semantic Web technologies within particular use cases.
The use cases are sometimes broad (which is understandable for a position paper). Some are not directly related specifically to the Semantic Web or semantic technologies (e.g., Maintenance and Replacement Engineering, Adaptation through optimization and reconfiguration), though the use of advanced knowledge and markup languages hints possible applications.

Overall, this is a well written paper, which has plenty of valuable, dense content on the topic.
Therefore I would recommend to accept the paper.
However, there is a potential problem: the paper is 3 pages too long and, to the best of my knowledge, the limit is 8 pages. If the paper is supposed to be shortened, then probably Section 2 might be shortened, as it is not always directly related to the Semantic Web, probably also Section 3 (though this would be a non-easy task, regarding the dense content).

Review #2
By Stefano Borgo submitted on 14/Sep/2019
Minor Revision
Review Comment:

This is a nice paper about the contribution of SW methodologies and technologies to manage the increasing complexity of cyber-physical systems (CPS) and to integrate the different perspectives (components, resources, subsystems, stakeholders) needed to design and operate CPS.
The paper exemplifies the problems with three use cases where SW technologies can not only improve information organization and management but also provide cross-domain interoperability.
The paper is clear in its goals, the domain terminology is well introduced, the use cases are well presented. It is a good contribution to the special issue.

One thing that is not clear in the introduction, and hampers the understanding of the nature of CPS, is that these systems are emerging from the tight integration of digital hardware, software and cyberware technologies (the latter being essentially about information, from which the core role of SW in CPS). The presentation talks about generic physical- and cyber-worlds and should explain what is the novelty of CPS with respect to traditional engineering systems.

There is another point that the authors can expand to strengthen the paper and its conclusions.
The authors suggest that the SW approaches in CPS may have a better impact if we stop applying them just within traditional boundaries (the discussed use cases provide examples of such subdomains: manufacturing, grid, smart building). The aim it to use SW and ontology to integrate (possibly all) the CPS aspects. This perspective applies the guiding viewpoint in the SW community to the specific domain and it is important to highlight this.
However, the fact that this view is not pursued enough today requires some analysis: why is it so? is it just a historical problem? is it due to today’s SW limitations?
My answer is that this integration cannot be achieved today because we still lack an ontological viewpoint on CPS. What are CPS ontologically? How can we ontologically integrate the different perspectives that CPS need? How can we integrate a part modelled as a component in one framework and as a subsystem in another?
We lack a conceptual framework, or even a shared guideline, in this domain. Without this I doubt we can achieve the needed integration discussed by the authors and I hope they will start a discussion of this problem towards the end of the paper.

Review #3
By Freddy Lecue submitted on 25/Sep/2019
Minor Revision
Review Comment:

The paper present future trends on semantic web technologies for Cyber-Physical Systems, which are systems relying on semantics for achieving complex tasks through various types of devices such as sensors, actuators.

I would add some initial reference to semantic (web) services which was one of a first attempt before IoT.

I would add more reference on smart cities infrastructures, which was quite high a few years ago.

Regarding smart building cf. work from IBM Research on connecting semantics with objects [1].

I would emphasize on the needs of light semantics, and use of knowledge graph as opposed to heavy ontology.

[1] Joern Ploennigs, Anika Schumann, Freddy Lécué: Adapting Semantic Sensor Networks for Smart Building Diagnosis. International Semantic Web Conference (2) 2014: 308-323