Construction is one of the most important industries worldwide, and its impact is significant in various aspects, such as environmental [3] or production costs [4]. The complexity of construction projects, the participation of numerous stakeholders, and the lack of transparency with data related to these projects contribute to increased costs and delays [5]. To address this matter, a potential approach consists of collecting valuable data associated with construction sites and analysing it to extract meaningful information that allows informative decision making that may reduce costs or delays [6].
Digital Twins (DTws) aim to address and solve these problems rooted in construction projects. DTws are a planning and optimisation approach that is based on simulations, allowing construction sites to improve the accuracy of their predictions, make more logical decisions, and develop well-informed planning processes [7]. Moreover, DTws possess two-fold characteristic: a physical aspect, which includes data pertaining to the actual state of affairs in the physical world, and a virtual aspect, encompassing the present operational status. The use of DTws in construction projects has been proved to increase operating efficiency by 35%, sustainability by 50%, reduce labour costs by 50%, increase productivity by 20%, and extend the space that can be used by 15% [8].
Even though DTws offer substantial advantages to construction projects, nowadays there is a lack of a standardised architecture for DTws that describes their design and guides their development and implementation. However, there are numerous DTw architectures in the literature [9], one of the most recognised and adopted is one that models DTws into three dimensions [10]: physical, virtual, and the connections between them. Several research proposals have refined these dimensions [11,12,13,14], nevertheless, only one has extended these dimensions, adding two new and promoting a five-dimensional architecture [15], which has become the reference architecture for DTws. However, such approach lacks formal semantics to describe the different dimensions and their features. This may lead to a set of problems to achieve semantic interoperability among DTws: i) there is no formal approach to discover the services offered by DTws and also to know to which dimension they belong; ii) there is no description of which model follows the data of a DTw; iii) there is no security specification for the dimensions of a DTw to address these problems. As a result, it is essential to develop a method for semantically describing the various aspects that model the five-dimensional architecture in a DTw.
In this article, an extension of the W3C standard Thing Descriptions (TD) ontology [16] proposed by the Web of Things group (WoT), called WoTDT (WoT Digital Twin ontology), is presented. The goal of WoTDT is to provide formal semantics to represent the five dimensions of a DTw and its features, allowing semantic discovery by reusing the TD ontology, and a common approach to describe the capabilities of a DTw allocated by dimensions. The WoTDT ontology was developed using the LOT methodology [17], following ontology engineering good practices. Finally, an evaluation of the ontology has been performed to ensure its quality by checking that it is pitfall-free and that it covers all the identified requirements.
The WoTDT ontology has been developed and used in the context of the European project COGITO. In this project WoTDT has been used to model the DTw of railway network construction site (with information such as workers, machines, IoT, . . . ). This ontology is public available as machine-readable format and as a human-readable document 1 . Furthermore, since the ontology is based on WoT, it allows i) to discover the services provided by the different dimensions [1]; ii) to facilitate the accessibility of information of a specific dimension [2]; iii) to define the security specification of each dimension using WoT capabilities by declaring a security scheme. Furthermore, WoTDT ontology can represent domain independent DTw, Digital Shadows or Digital Models, and it is meant to be extended, when required, with domain specific terms (such as sensors, building parts or machinery). Practitioners shall use the WoTDT ontology as a common data model for interoperability within their infrastructure, and extend it with specific terms from their infrastructure domain to increase the expressiveness of the ontology.
The rest of this article is structured as follows. Section 2, provides an overview of the literature and research proposals related to DTws in terms of architectures, ontologies, and WoT. Section 3 includes the development of the WoTDT ontology using the LOT methodology. Section 4 provides a use case of this ontology in the COGITO project. Finally, Section 5 recaps the conclusions of this research.
The concept of "Digital Twin" was coined by Michael Grieves in the field of Product Lifecycle Management (PLM) [10], defining it as a virtual representation of a physical object or system that can be used to analyse its behaviour and performance. The proposal models an architecture of a DTw that consists of three dimensions: the physical, which represents the actual asset in the real world; the virtual, which represents the software representation of the physical asset; and the connection parts, which exist between the physical and virtual dimensions.
After the three-dimensional approach, the researchers focused on optimising aspects of these dimensions to tackle new challenges [11,12,13,14]. For example, NASA [12] improved existing simulations performed in the virtual dimension with the goal of improving the management of US Air Force aircraft throughout their lifetime by creating individualised structural management plans, and in the physical dimension by including real-time data collection.
In 2017, Tao and Zhang [15] proposed a five-dimensional domain independent architecture model of a DT, extended from the three-dimensional approach, that provided theoretical guidance for the digitalisation and intellectualisation of the manufacturing industry. Also, they adapted the five-dimensional approach to specific domains, such as prognostics and health management domain [18]. In this proposal, the physical dimension of Grieves' proposal is called the Physical Entity Dimension, and the connections are called Digital Twin Connection Dimension. The virtual dimension of Grieves' proposal is divided into three dimensions: the Virtual Entity Dimension, which includes the description of the DTw models (such as rules, behavioural, physical, and geometric models); the Digital Twin Data Dimension, which stores all the data used by the DT; and the Digital Twin Service Dimension, which includes all the services used by the DTw.
The goal of the WoTDT ontology is to model this five-dimensional architecture. Therefore, DTw proposals in the literature that rely on ontologies to model this architecture, or similar ones, and/or apply WoT in the context of DTws are analysed.
Ontologies in DTws: The application of Semantic Web technologies in the creation, management, and modelling of DTws in industry has grown in recent years. In particular, guidelines have been proposed on how to incorporate ontologies and their benefits for DTws [19]. Using these ontologies has one of these two main goals: i) to improve the visualisation and understanding of DTw data [20], or ii) to describe the architecture model of a DT [21], which is the goal of the WoTDT ontology.
In the context of the second goal there is a study proposed by Sumit Singh et al. [22], where an ontology is developed to capture the domain knowledge and maintain the semantics of the asset functions and basic characteristics during its operational phase. Another proposal by Charles Steinmetz et al. [21] conducted a study that expands the IoT-Lite ontology [23] by incorporating classes associated with the virtual dimension of the three-dimensional model, allowing external applications and systems to access the virtual dimension with a particular protocol. In addition, Marah and Challenger [24], introduce a definition of an ontology for agent-based digital twins for Cyber-Physical Systems. Furthermore, Meijers from Microsoft Azure introduced the Digital Twins Definition Language (DTDL) [25] based on an RDF metamodel to represent DTws within their proprietary tools.
WoT in DTws: In the literature, WoT has not been used to model the dimensions of a DTw. However, the closest proposal is the research presented by Ricci et al. [26], where a description of an ecosystem of DTws has been developed using semantic web technologies in the healthcare domain. In their research, the combination of DTw and WoT allows greater precision in monitoring construction assets in real time, interoperability with other systems, and better behaviour and performance over time. Another research, presented by Pittaras et al. analysed how to combine the WoT standard with smart contracts to describe DTws of physical devices [27]. In addition, in the research conducted by Luca Bedogni and Federico Chiariotti [28], an architecture based on the WoT standard is developed to provide a DTw of a smart environment, such as sensors used in a construction-related project. Finally, Xiaochen Zheng et al. in [29] bring together these technologies, where WoT is applied in Cognitive DTws (CDTws) with the intention of standardising and achieving data interoperability.
This article introduces the first proposal, as far as we know, that covers all five dimensions of a DTw and their characteristics. Moreover, since the ontology has been developed as an extension of WoT, it will provide advantages such as enabling the discovery of services across dimensions, or facilitating the access to the information from a particular dimension, thereby promoting data interoperability.
In this article, a novel approach is presented to semantically represent Digital Twins (DTw) that follow the five-dimensional architecture along with their characteristics and functionalities. To this end, an extension of the W3C standard Thing Description (TD) ontology published by the Web of Things (WoT) group has been developed. This novel ontology, named WoTDT provides an approach that describes all the dimensions and internal capabilities of a DTw, and ranges from related metadata to the functions it offers using TDs enriched with this ontology.
The WoTDT ontology has been developed following the LOT (Linked Open Terms) methodology [17]. This methodology is based on the ontological engineering activities described in the NeOn methodology [30]. The LOT methodology outlines a process that includes the following activities: i) specifying ontological requirements; ii) implementing the ontology; iii) publishing the ontology; and iv) maintaining the ontology. The following subsections provide further details about these activities and their implementation to deliver the WoTDT ontology.
The specification of the ontology requirements is derived from the analysis of the proposed fivedimensional model and refined during its adoption and use in the COGITO H2020 project. From these inputs, a collection of ontological requirements was generated in the form of competency questions and sentences in natural language. Table 1 shows the fifteen requirements obtained from the specification process.
To streamline the implementation process, a conceptualisation of the ontology was proposed using the requirements identified in the previous step as input. This conceptualisation incorporates the key concepts to represent the five dimensions when applied to the construction domain. Figure 1 shows the representation of this conceptualisation, which follows the Chowlk Visual Notation [31] that establishes a connection between the stereotypes used in the profile with OWL, RDF (S) constructs and certain OWL 2 constructs. The use of prefixes indicates the ontologies in which each concept or relation is defined. For example, dt:DigitalTwin is defined in the "https://w3id.org/def/digitaltwin#" namespace. Furthermore, this ontology reuses the core ontology of SAREF, such as saref:Device; the DCAT ontology (Data Catalog Vocabulary), such as dcat:Resource; the HCTL ontology (WoT Hypermedia Controls Ontology), such as hctl:hasTarget; apart from the WoT Thing Description ontology. As shown in Figure 1, the class of dt:DigitalTwin is represented as a rdfs:subClassOf of td:Thing, allowing inheritance of properties contained in the td:Thing class. In addition, the different dimensions are conceptualised using the classes: dt:PhysicalEntity, dt:VirtualEntity, dt:DigitalTwinData, dt:DigitalTwinService and dt:DigitalTwinConnection. These five-dimensional conceptualisations will be explained separately in the following subsections.
Physical Entity Dimension: In the WoTDT ontology, the dt:PhysicalEntity class represents the dimension of the physical entity, defined as a set of various subsystems, objects, and sensors. devices. These subsystems can include dynamic systems, control systems, maintenance systems, among others, and can be combined for a specific task [15]. To represent all these subsystems, objects, and sensor devices, the class dt:Component has been created. The dt:Component class can be shown in Figure 2 where it contains subclasses such as dt:MainComponent, that represents the main component of the dimension of the physical entity in the case of multiple; and saref:Device, that represents devices such as sensors (saref:Sensor) and actuators (saref:Actuator), allowing this dimension to read and act on the physical asset of the DTw.
Virtual Entity Dimension: For the dimension of the virtual entity, in the WoTDT ontology, this dimension is represented in the class dt:VirtualEntity, defined as a set of various data models that represent all the information allocated in the digital twin [15]. To represent the models allocated in this dimension, the class dt:Model has been created, defined as a representation or conceptualisation of the data registered in the dimension of the virtual entity. Following the same approach as in the previous subsection, in Figure 3 are represented the different subclasses of the class dt:Model, containing from non-semantic models such as dt:RulesModel, dt:BehavioralModel, dt:PhysicalModel and dt:GeometricModel described in [15]; and semantic models such as dt:OntologyModel, dt:MappingModel and ShapesModel. Furthermore, as shown in Figure 1, each of these dt:Model, has a specific format, represented by the class dt:Format, containing attributes (owl:DatatypeProperty) such as the route where the model is stored (hctl:hasTarget), and the extension of the model (dt:hasExtension), that can be for example in the case of a dt:OntologyModel the "ttl" extension that refers to Turtle serialisation. Digital Twin Data Dimension: Once the physical and virtual entities have been explained, the dimension of digital twin data will follow. In the WoTDT ontology, the dimension of the digital twin data is represented by the class dt:DigitalTwinData, defined as the dimension of the digital twin where all data are contained and used by the DTw [15]. As shown in Figure 1, these data are represented with the class dcat:Resource. In this case, we reuse the DCAT ontology instead of the WoT TD ontology to represent the data, because DCAT is more complete and allows describing a larger volume of data (as used in DTws) with a larger number of features, such as dcat:Dataset, dcat:DataService or dcat:Catalog. This can be useful in the case that there are multiple volumes of data categorised by semantic and non-semantic.
The following dimension depicted is the dimension of digital twin services. This dimension is represented with the class dt:DigitalTwinService, defined as the dimension of the DTw that contains all the services that are used to perform all the processes of the digital twin [15]. As explained earlier, WoT allows us to de-scribe the services of a DTw, this is the reason why we reuse in this dimension the class td:interactionAffordance to represent the different existing services in the DTw, under the class of dt:DigitalTwinService. Furthermore, this class allows to show the possible choices to the DTw consumers, suggesting how they can interact with the DTw services.
Digital Twin Connection Dimension: The last dimension is the digital twin connections, which are represented by the class dt:DigitalTwinConnection. This class is defined as the dimension of the DTw that contains all the connections that exist between the different dimensions of the DTws and the information used [15]. To represent these connections, the class dt:Connection has been created and defines the existing connection between different dimensions. The dt:ConnectionPoint class is responsible for representing a particular endpoint of a dimension that functions as a provider or consumer of information (represented by dt:hasProvider and dt:hasConsumer) within the connection. With this approach, there will always be a connection with only two connection points. The different types of dt:ConnectionPoint are represented in Figure 4. The authors decided to introduce the class dt:ConnectionPoint instead of using the existing SEAS ontology [32]. This decision was made because SEAS only considers connections between systems, whereas in the context of DTws, connections may not always be limited to systems. In addition, even though the ontology has been developed within the construction field, it is extensible to other domains, as it allows for the inclusion of new subclasses within the various classes of dimensions to address the requirements of a specific domain. Once the WoTDT conceptualisation was defined, it was encoded in OWL using Chowlk [33] and Protégé2 , and stored in a GitHub repository 3 .
The evaluation to identify typical errors during ontology implementation has been carried out using the OOPS! tool [34]. Several important and minor pitfalls have been identified. However, these important pitfalls do not affect the consistency, reasoning, or applicability of the ontology. In addition, when it comes to minor pitfalls, they mainly involve identified unconnected elements. The remaining errors identified by the tool were corrected accordingly.
Furthermore, the authors performed a systematic analysis to ensure that the ontology satisfies the requirements elicited in Table 1. To assist in this validation, a set of tests 4 was defined and executed using the Themis tool [35], which implements the testing methodology described in [36]. After executing the tests, it was discovered that all of them were successfully completed, indicating that the developed ontology meets all the defined requirements.
Once the ontology is developed and evaluated, it has to be published online. To achieve this goal, OnToology [37], a web-based system, is used. OnToology is designed to work seamlessly with Git-based environments and incorporates various pre-existing documentation tools. In addition, OnToology offers two options for publication using content negotiation mechanisms. The first option is to publish the ontology using a permanent id through the services provided by "https://w3id.org". The second option is to download a bundle containing all the necessary files to publish the ontology on a server. For the WoTDT ontology, the first option was selected, publishing it under the URI "https://w3id.org/def/digitaltwin", where the ontology is available in machine-readable format and as a human-readable document.
Finally, to support the maintenance activity in the developed ontology, an issue tracker5 is used. Hence, for users, domain experts, or ontology developers to suggest new requirements, identify bugs, or provide any suggestions, they are required to create an issue in the repository. This issue tracker enables the monitoring of all the issues suggested, and once an issue is open, the ontology development team needs to discuss and make a decision on whether the proposal presented in the issue should be implemented in the ontology or rejected.
This section provides an example of an application of the WoTDT ontology in the European H2020 construction-related project COGITO. In particular, the section aims to show a DTw description using this ontology, i.e., data instantiation, for a wall in a pilot COGITO construction site.
The instantiation of the data in the given example is obtained from the URI used to publish the different resources used in COGITO (https://data.cogito.iot.linkeddata.es/resources/<prefix>/<id>). For instance for accessing the resource sdt:01U2O the public url is the following: https://data.cogito.iot.linkeddata.es/resources/sdt/01U2O. In Figure 5, an instantiation of a DTw and its respective dimensions is shown. After the DTw has been instantiated, an example of how each dimension can be instantiated has been provided.
First, in Figure 6, an example of the Physical Entity dimension instantiation is presented. In this example, a "Wall" is represented as a dt:Component, that also is a dt:MainComponent and a facility:Element.
Second, in Figure 7, an example of Virtual Entity dimension instantiation is given. In this case, the dimension of the virtual entity has a model that is a dt:OntologyModel, found in four different dt:Format, containing each of them the specific target URI to get the model.
Third, in Figure 8, an example of Digital Twin Data dimension is shown. In this dimension, the instantiation of the data resources is presented, where a dcat:Dataset is representing the Knowedge Graph (KG) [38] of the DTw. Furthermore, the representation of access to the dcat:Distribution and dcat:DataService are provided, in order to retrieve the KG data.
In fourth place, in Figure 9, an example of instantiation of the dimension of the Digital Twin Service is presented. In this particular case, the various services that offer the DTw are depicted as td:PropertyAffordance and td:ActionAffordance. These services provide their respective hctl:Form, which specifies the URL that the consumer needs to access the service. It is important to remark that WoT allows describing data resources using the TD standard [2]. Also, promotes a standardised discovery of these resources by registering their TDs into WoT directories and finding them using syntactic and semantic queries [1]. The expressiveness of TDs allows describing different features of the data resources; how they are accessed, the protocols they use, their security methods, and metadata identifying the real world entity they represent. For instance, a TD could contain metadata describing a building, specifing its HTTP data endpoints and how to invoke them, also, the kind of data they provide (such as columns position, light bulbs status, etc), and a basic auth mechanisms to access.
Due to the fact that WoTDT ontology inherits from TD it also acquires all its benefits, including expressiveness and the discovery capability in WoT directories. This ontology aims at enhancing the expressiveness of TDs with domain specific terms for DTws, an expressivness that TDs lack due to its domain neutrality.
As a running example, Figures 5-10 show a TD extended with the WoTDT ontology. As a result the TD describes the five dimensions of a DTw (domain specific terms) and also describes the data resources that conform the DTw (neutral domain terms). Note that these examples do not show explicitly security schemes but they could be specified. Also, the TD of these examples is compliant with the WoT discovery standard, and therefore, it could be registered and found in a WoT directory.
DTws in the construction domain allow for the accuracy of needed predictions to be improved, making more logical decisions, and developing well-informed plans. However, there is no standardised architecture that provides a precise definition of a DTw and offers guidance for its development. In the literature, there are research proposals that model their architecture in a three-dimensional model [14]. Subsequently, a novel proposal presented an extended version that models the DTw in five dimensions [15], providing a more complete understanding of the architectures of the DTws. However, DTws lack formal semantics, which hinders interoperability with third-party DTws or external services.
To address this problem, the WoTDT ontology is developed as an extension of the Thing Description ontology proposed by the W3C Web of Things group, which is a standard that allows the description of services, promoting interoperability, discoverability, accessibility, security and scalability. As a result, this ontology will enable a more precise comprehension of the DTws and will provide direct access to all the system functionalities for the described DTw.
Moreover, despite being initially designed for the construction domain, the ontology is flexible enough to accommodate the incorporation of additional subclasses into classes related to different dimensions. This capability enables the specific requirements of a particular domain to be addressed. However, some issues may arise when adding these new requirements that may trigger changes in the ontology. Therefore, a potential area for future research involves expanding the ontology to encompass a wide range of information from various domains. Furthermore, there might be value in representing the aggregation of an ecosystem of DTws, as it allows for the conceptualisation of the interconnections among various DTws.
This work has been supported by COGITO funded by the European Union's Horizon 2020 research and innovation programme under grant agreement no. 958310 and by the Madrid Government (Comunidad de Madrid-Spain) under the Multiannual Agreement with the Universidad Politécnica de Madrid in the Excellence Programme for University Teaching Staff, in the context of the V PRICIT (Regional Programme of Research and Technological Innovation).