Process-aware IIoT Knowledge Graph: A semantic model for Industrial IoT integration and analytics
Introduction
Industry 4.0 represents the fourth industrial revolution and is characterized by the introduction of digital and other innovative technologies in manufacturing. One of the key enabling technologies is the Internet of Things (IoT), defined as a network of interconnected devices that collect and exchange data through the Internet [1]. IoT encompasses a wide variety of application domains, with different and contrasting requirements [2], [3]. Focusing more specifically on industrial applications, the term Industrial Internet of Things (IIoT) has been coined. In contrast with other applications, where humans have a central role (connected devices are consumer electronic devices, and applications are devoted to improve human awareness of the surrounding environment) [4], IoT in industry is mainly devoted to connect machines to each other and to control systems, in order to enable self-organization, self-optimization, self-healing, towards a more autonomous and intelligent manufacturing system. In this respect, the focus is on the production line. Information is sensed and exploited on-line and locally in order to react to events, or exchanged with few peers for coordination or remote control during manufacturing. More recently, however, larger integration scenarios have been envisaged, up to the level of the whole factory, or even the enterprise, for collecting and analysing flows of data enabling optimization and planning. Correspondingly, the reference architecture is emerging as an edge/fog-cloud architecture [5], [6].
Integration at cloud level of the huge data streams coming from the field, and providing a uniform view of the course of events are major issues. In particular, a relevant challenge can be summarized by the following research question: how to bridge the abstraction gap between raw sensor data and higher-level knowledge in the organization?
It is recognized that model-based approaches may support the integration effort providing fundamental reference information. In particular, processes are recognized as a major means of integration in an organization. Processes are a set of interrelated activities, aimed at realizing a product or service that contributes to the achievement of the organization’s goal. In the development of interrelated activities, many business units are involved. Operating on the same process, business units’ resources must be integrated and coordinated for achieving the objective of the process. In this sense, the process is therefore an element of homogenization of the organization. Recently, the work in [7] supports this view, enlightening the fact that process analytics, execution, and monitoring based on IoT data can enable an even more comprehensive view of systems and realize unused potential for optimization, provided that some challenges are tackled and resolved.
In the present paper, we aim at answering to the research question by providing a process-centric semantic model that formalizes and relates the fundamental elements involved in IIoT for cloud-level integration and analytics. We recognize these elements in (i) sensors, (ii) processes and (iii) related performance indicators, and (iv) the factory, intended as machinery and the environment that contains it. A comprehensive ontology is built following state-of-the-art design principles, in particular reuse, and abstraction.
The ontology acts as a conceptual knowledge layer providing the structure of the Process-aware IIoT Knowledge Graph, that includes all instances of elements listed before. In particular, it enables the enrichment of raw sensor data with information about process activities and the physical production environment and, as such, their contextualization. The proposal is enriched with a framework for querying the Knowledge Graph, whose capabilities are demonstrated by considering the production of metal accessories as case study.
The rest of the paper is organized as follows: Section 2 discusses the related literature. Section 3 introduces the case study. Then, Section 4 discusses the layered-based principle adopted to organize the knowledge artifacts, while Section 5 describes the proposed ontology, illustrating the component ontologies and the integration strategy. Sections 6 Process-aware IIoT Knowledge Graph construction, 7 Graph exploration and analysis are respectively devoted to discuss the construction and exploitation of the Process-aware IIoT Knowledge Graph. Finally, Section 8 draws some conclusions and future work.
Section snippets
Related work
Interoperability is a key challenge in IIoT [8], due to the presence of multiple and heterogeneous data sources using different knowledge representations. In this context, ontologies are seen as a promising tool to solve interoperability problems [9], as they provide commonly agreed data models that are understandable by both humans and machines. In the field of IIoT, in particular, ontologies may support a wide range of activities (e.g., design, simulation, planning and scheduling, performance
Case study
This section is devoted at presenting a case study that will be used as an illustrative example throughout the work. The proposed case study is a mid-size manufacturing company that produces metal accessories, which are obtained through fabrication and assembly of several metal parts. Such accessories are highly customizable and are produced in different sizes, shapes and colours, which implies that different products may be subject to different processing phases. The company has an industrial
Layered framework
In this section we discuss how information (or “knowledge artifacts”) in the IIoT is categorized based on its typology and the object it represents.
Firstly, we recognize two main perspectives that are intertwined in the IoT environment, namely focusing on processes and sensors:
- •
the process perspective focuses on the definition, planning and execution of business processes, including all the process-related information;
- •
the sensor perspective focuses on the characterization of sensors and their
A semantic model for IIoT
This section is devoted to discuss the model proposed for Industrial IoT. The model is able to completely represent the knowledge artifacts introduced in the previous section and highlight the relations among them. The model is shown in Fig. 3. It is an ontology which reuses and integrates, according to the best practices in ontology engineering, a number of existing ontological models, as well as a novel ontology for the representation of process models specifically built for the purposes of
Process-aware IIoT Knowledge Graph construction
The ontological model discussed in the previous section represents the vocabulary for the definition of the Process-aware IIoT Knowledge Graph, which is represented as an RDF graph including all the knowledge artifacts belonging to the layers from the perspectives introduced in Section 4. While some of the artifacts, e.g., the process model, can be directly mapped to corresponding classes and properties in the ontological schema, some others, such as the event log and the observations, require
Graph exploration and analysis
In this section, we discuss how the model can be exploited for querying and explorating the Process-aware IIoT Knowledge Graph. The functionalities hereby described are aimed to define the query for extracting and analysing data under a set of user conditions.
The graph includes two different types of quantitative measures that can be analysed, namely those produced by sensors, which are described through instances of ObservableProperty (e.g., “Temperature”), and those attached to tasks and
Conclusions and future work
The paper discussed a semantic approach for the integration of data streams coming from the field in the context of Industrial Internet of Things, providing a more comprehensive view of systems and supporting analytic tasks. The novelty of the approach is given by the process perspective taken in the conceptualization of the ontology and the related Process-aware IIoT Knowledge Graph, which allows to bridge the abstraction gap between raw sensor data and higher-level knowledge in a
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Claudia Diamantini Ph.D. (member IEEE, senior member ACM) is Full Professor at Dipartimento di Ingegneria dell’Informazione, Università Politecnica delle Marche, where she also holds the role of Vice Dean of the Faculty of Engineering. Her research interests include data mining and knowledge discovery, process modelling and mining, data semantics and knowledge graphs. On these topics she has worked within national and international projects, and authored more than 170 publications
References (50)
- et al.
Internet of Things (IoT): A vision, architectural elements, and future directions
Future Gener. Comput. Syst.
(2013) - et al.
The SSN ontology of the W3C semantic sensor network incubator group
J. Web Semant.
(2012) - et al.
SOSA: A lightweight ontology for sensors, observations, samples, and actuators
J. Web Semant.
(2019) - et al.
An ontology-based framework to support intelligent data analysis of sensor measurements
Expert Syst. Appl.
(2014) - et al.
Abnormal situations interpretation in industry 4.0 using stream reasoning
Procedia Comput. Sci.
(2019) - et al.
Ontology network to support the integration of planning and scheduling activities in batch process industries
J. Ind. Inf. Integr.
(2022) - et al.
Multi-factory production planning using edge computing and IIoT platforms
J. Syst. Softw.
(2021) - et al.
SemPI: A semantic framework for the collaborative construction and maintenance of a shared dictionary of performance indicators
Future Gener. Comput. Syst.
(2016) - et al.
Building instance graphs for highly variable processes
Expert Syst. Appl.
(2016) - et al.
A novel IoT platform for the era of connected cars
Extraction of user daily behavior from home sensors through process discovery
IEEE Internet Things J.
Industrial Internet of Things: Challenges, opportunities, and directions
IEEE Trans. Ind. Inf.
Fog computing conceptual model
The Internet of Things meets business process management: a manifesto
IEEE Syst. Man Cybern. Mag.
A comprehensive survey on semantic interoperability for Internet of Things: State-of-the-art and research challenges
Trans. Emerg. Telecommun. Technol.
A survey of data semantization in Internet of Things
Sensors
Knowledge graphs
Synth. Lect. Data, Semant. Knowl.
A literature review of sensor ontologies for manufacturing applications
A novel ontology for sensor networks data
Building a sensor ontology: A practical approach leveraging ISO and OGC models
OWL web ontology language overview
W3C Recomm.
IoT-lite: a lightweight semantic model for the Internet of Things
Cited by (8)
Semantic modeling and design patterns for IoT ecosystems
2023, Future Generation Computer SystemsCitation Excerpt :To validate the model, a real-life telecommunication dataset consisting of 7.2M Internet access trace data gathered from more than 3k base stations and referring to nearly 10k mobile users is exploited, so to enable the comparison between the proposed approach and a set of baseline alternative models. The article “Process-aware IIoT Knowledge Graph: a semantic model for Industry IoT integration and analytics”, authored by C. Diamantini, A. Mircoli, D. Potena, and E. Storti [2], addresses the important role that semantic models can play in the Industry 4.0 context when the integration of significant data streams from IIoT sources is required at cloud level and when data analytics are necessary. Starting from the ontological description of IIoT sensors, processes, and KPIs, the authors propose a process-aware knowledge graph aimed at enriching raw sensor data with information coming from two complementary perspectives: the one describing process activities and the one associated with the physical production environment where the IIoT sensors are deployed.
On the Interplay Between Business Process Management and Internet-of-Things: A Systematic Literature Review
2024, Business and Information Systems EngineeringA kind of intelligent dynamic industrial event knowledge graph and its application in process stability evaluation
2024, Journal of Intelligent ManufacturingAn IoT architecture to integrate different machine tools into a compound OPC UA interface
2023, IT - Information TechnologyA Graph-Based Approach for Representing Water Treatment Process Knowledge
2023, 14th International Conference on Information, Intelligence, Systems and Applications, IISA 2023
Claudia Diamantini Ph.D. (member IEEE, senior member ACM) is Full Professor at Dipartimento di Ingegneria dell’Informazione, Università Politecnica delle Marche, where she also holds the role of Vice Dean of the Faculty of Engineering. Her research interests include data mining and knowledge discovery, process modelling and mining, data semantics and knowledge graphs. On these topics she has worked within national and international projects, and authored more than 170 publications
Alex Mircoli received a Ph.D. degree in Information Engineering from Università Politecnica delle Marche, in 2019, with a thesis entitled “Lexicon- and learning-based techniques for emotion recognition in social contents”. Currently, he is a PostDoc at Dipartimento di Ingegneria dell’Informazione, Università Politecnica delle Marche.
Domenico Potena received the Ph.D. in Information Systems Engineering from Università Politecnica delle Marche, Italy, in 2004. At present, he is an associate professor at Dipartimento di Ingegneria dell’Informazione, Università Politecnica delle Marche. His research interests include process mining, knowledge discovery in databases, data mining, data warehousing, information systems and service oriented architectures.
Emanuele Storti received the Ph.D. degree in Computer Engineering from Università Politecnica delle Marche in 2012 and is currently an assistant professor at Dipartimento di Ingegneria dell’Informazione. His research interests include knowledge graphs, knowledge management, data integration.