Special Issue on Semantic Web for Industrial Engineering: Research and Applications

1.Introduction

Semantic Web languages, models, and technologies are increasingly finding widespread research and practical applications within the domain of Industrial Engineering. The key motivations behind applications of Semantic Web in industrial domains is twofold. On one hand, the extensive utilization of computer-based technologies in manufacturing organizations raises the need for rigorous approaches for managing and sharing data. On the other hand, achieving interoperability among different systems and stakeholders requires alignment between different terminologies. Specifically, in its current state, Semantic Web approaches within industrial domains address challenges that can be broadly classified into five categories:

Various initiatives by national and international organizations have been initiated to disseminate and facilitate the utilization of the SW for industrial requirements. These initiatives aim to collect, enhance, and promote existing methodologies, best practices, ontologies and tools, and foster their further development and application. Examples of these initiatives are the European project OntoCommons11 (part of the Horizon 2020 research and innovation program), the global endeavors of the Industrial Ontologies Foundry (IOF),22 the W3C Linked Building Data Community Group,33 workshop series like FOMI (Formal Ontologies Meet Industry)44 and LDAC (Linked Data in Architecture and Construction)55 as well as digital twin-related initiatives like the UK National Digital Twin Programme (NDTP)66 and the Building Digital Twin Association (BDTA).77

2.Special issue topics

The purpose of this special issue, titled “Semantic Web for Industrial Engineering: Research and Applications”, is to collect contributions focusing on the application of Semantic Web in various industrial engineering settings such as discrete manufacturing (e.g., aerospace, automotive, machinery, electronics), continuous production (e.g., chemical engineering, oil and gas industry), product design, Architecture, Engineering and Construction (AEC). The topics we aimed to cover in this special issue include the following ones:

3.Contributions

Initially, 18 papers were submitted to the Special Issue, of which 12 have been accepted for publication. Herein, we present a summary of the accepted papers. Furthermore, Table 1 shows the mapping of the articles against the challenges (Section 1) and topics (Section 2).

Bareedu et al. [2] address challenges related to guidelines for data modeling provided in industrial standards but in unstructured formats. The use of informal specifications hinders the automated validation of data models reliant on these standards, potentially leading to costly interoperability errors. The paper proposes an approach to extract and represent modeling constraints from industrial standards as machine-actionable rules, particularly in the context of the OPC UA standard. The authors demonstrate the feasibility of this approach by formally representing modeling constraints using Semantic Web languages, successfully identifying constraints within specification documents and achieving the semiautomatic translation of the constraints into formal means.

Compagno and Borgo [5] aim to contribute to the establishment of a comprehensive ontological modeling of the engineering concepts of functionality, behavior, capability, and capacity, including the relationships between them. Using the foundational ontology DOLCE as top-level modeling framework, the paper conducts a conceptual analysis of these notions and formalizes the resulting logical theory in first-order logic and OWL. The study emphasizes differentiating functions from their implementation methods, distinguishing between capabilities and capacities of a product, and elucidating their relationships within engineering systems.

Donkers et al. [7] propose a scalable method for acquiring continuous occupant feedback, integrating it with building information using Semantic Web technologies. The Occupant Feedback Ontology (OFO) facilitates semantic description of feedback. The Mintal smartwatch app collects location, medical data, and survey responses, applying the ontology for integration with linked building data. A case study demonstrates the creation of a semantic digital twin, enhancing occupant-centric decision support tools and offering insights into perceived comfort levels.

Giustozzi et al. [9] introduce a semantic framework for evolving knowledge bases related to condition monitoring in the scope of Industry 4.0. The COInd4 ontology represents manufacturing domain resources, processes, and contextual information. Stream reasoning integrates sensor data with contextual knowledge, detecting situations leading to potential failures in real-time. A lattice structure organizes these situations by severity, providing a roadmap for decision support. The approach ensures adaptive process behavior without interrupting manufacturing processes.

Lambrix et al. [11] tackle the challenge of accessing and integrating heterogeneous data stored in various computational materials databases. The paper introduces the Materials Design Ontology, inspired by the OPTIMADE effort, aiming to enhance interoperability among multiple databases in computational materials science. The paper details the ontology and showcases its use in a proof-of-concept implementation for data access and information systems interoperability.

Stüber and Frey [20] explore the use of Semantic Web technologies to enhance the FAIRness of Modeling and Simulation (M&S) capabilities in industrial systems. In particular, it presents an open-source proof-of-concept implementation based on the Functional Mock-up Interface (FMI) standard. This implementation exposes models, instances, and simulation results through a hypermedia API, demonstrating increased FAIRness and support for use in loosely coupled systems.

Umbrico et al. [24] present an extension of the Sharework Ontology for Human-Robot Collaboration (SOHO) to enhance the characterization of behavioral constraints in collaborative tasks. The work introduces a knowledge extraction procedure for automating the synthesis of Artificial Intelligence plan-based controllers. The generality of the ontological model and the synthesized planning domains are validated in real industrial scenarios with collaborative robots.

Werbrouck et al. [26] introduce ConSolid, a decentralized ecosystem designed for multi-stakeholder collaborations dealing with heterogeneous data in industries like construction. Leveraging Solid specifications for Web decentralization, ConSolid extends these specifications to facilitate semantic interoperability among diverse subdomains. The ecosystem employs metadata-generated virtual views via SPARQL interfaces, enabling data aggregation and alignment across multiple vaults in a scalable manner.

Woods et al. [27] study maintenance for asset-intensive organizations by focusing on data quality of Maintenance Work Order (MWO) records. The authors present a reference ontology for maintenance activity terms derived from natural language processing of MWOs, and an application-level ontology demonstrating the use of the reference ontology in an industrial use case. The end-to-end NLP-ontology pipeline identifies data quality issues and infers relevant activity classes, offering a practical approach for organizations to enhance maintenance work management processes through ontological workflows.

Yahya et al. [28] focus on developing a benchmark dataset for generating Knowledge Graphs (KGs) in Industry 4.0 production lines. It highlights the benefits of utilizing ontologies and semantic annotations to demonstrate the advantages of KGs in the Industry 4.0 domain. Collaboration with industry professionals has resulted in a KG based on the Reference Generalized Ontological Model (RGOM), showcasing adaptability and usefulness in supporting decision-making processes within production lines.

Yaman et al. [30] apply semantic web standards to enhance data quality governance in the architectural, engineering, and construction (AEC) domain, focusing on Ordnance Survey Ireland (OSi). The approach establishes a unified knowledge graph for data quality measurements across a complex, heterogeneous data production pipeline. It formalizes mappings between semantic models of data quality dimensions from ISO and W3C standards, enabling end-to-end data quality reporting and analysis. The use of a knowledge graph and semantic web standards unifies distributed data quality monitoring, offering a standards-agnostic end-to-end data dashboard for OSi’s data publishing pipeline.

Zhao et al. [33] address the need for seamless information exchange when dealing with dynamically changing 3D environments in shopfloor and warehouse robotics. The authors propose an ontology model integrating contextual, topologic, and geometric information for both rigid bodies and free space. The evolvable knowledge model supports simulated task-related information, improving interoperability for path planning systems. The ontology allows for updating contextual semantics related to specific applications while maintaining geometric and topological models through semantic links.

4.Remarks

The papers featured in this special issue, along with the referenced international research and application works, underscore the considerable interest of the industrial engineering community in ontologies and Semantic Web technologies. These technologies play a crucial role in fostering robust data management practices within organizations, offering valuable support to experts and stakeholders in effectively handling large and heterogeneous datasets in their daily routines.

At the same time, as researchers in ontology engineering, we often face criticism asserting that the efforts invested in the development of ontology-based models and technologies may not be justified, with alternative approaches being deemed more readily exploitable and maintainable. Additionally, despite the initial introduction of ontologies as a common “interlingua” across systems to address knowledge portability and counteract the proliferation of data model, we constantly witness the emergence of new ontologies. Another prevalent critique centers around the challenge of determining when one has created a “good” ontology, given the absence of standardized benchmarks in the ontology evaluation process.

It is essential to acknowledge that ontology engineering, as a scientific endeavor, is not a magical solution. The research community must enhance its practices by consistently evolving methodologies to (a) aid stakeholders in developing ontology-based systems for their businesses; (b) build and maintain Knowledge Graphs in sync with the ontology evolution; (c) establish standardized benchmarks for systematic ontology testing; and (d) define ontology governance models for different types of organizations. Moreover, there is a duty to educate future generations of computer scientists and engineers, providing them with the interdisciplinary skills needed for ontology design, usage, and maintenance. Given that ontologies are knowledge-based models, it is unavoidable that their representation will entail some complexity when the human knowledge they aim to represent exhibits a certain degree of functional complexity for the intended application. Finally, despite encountering scenarios where various ontologies coexist, even within specific application domains, a straightforward solution, in our opinion, remains elusive. Efforts like OntoCommons or the IOF, mentioned earlier, rely either on formal mappings to align ontologies to ensure a certain level of interoperability among the data modeled on their basis, or develop reference ontologies for data management practices. The first approach promotes a plurality of perspectives about the engineering domain, while the second one relies on single, monolithic models to be widely adopted within communities.

While acknowledging that ontologies may not guarantee complete information systems interoperability, they can, at the very least, be systematically used to discern semantic convergences and divergences in data. In a nutshell, the maturity of the discipline and the business exploitation of ontology-based tools in the (near) future hinge on the research community’s efforts and its ability to meet the evolving requirements of society.