Development of Safety Domain Ontology Knowledge Base for Fall Accidents

Abstract

Extensive research in the field of construction safety has predominantly focused on identifying the causes and impacts of construction accidents, evaluating safety plans, assessing the effectiveness of safety education materials, and analyzing relevant policies. However, comparatively limited attention has been given to the systematic formation, management, and utilization of safety-related information and knowledge. Despite significant advancements in information and knowledge management technologies across the architecture, engineering, and construction (AEC) industries, their application in construction safety remains underdeveloped. This study addresses this gap by proposing a novel ontology-based framework specifically designed for construction safety management. Unlike previous models, the proposed ontology integrates diverse safety regulations and terminologies into a unified and semantically structured knowledge model. It comprises three primary superclasses covering key areas of construction safety, with an initial focus on fall hazards—one of the most frequent and severe risks, particularly in roofing activities. This domain-specific approach not only improves semantic clarity and standardization but also enhances reusability and extensibility for other risk domains. The ontology was developed using established methodologies and validated through reasoning tools and competency questions. By providing a formally structured, logic-driven knowledge base, the model supports automated safety reasoning, facilitates communication among stakeholders, and lays the foundation for future intelligent safety management systems in construction. This research contributes a validated, extensible, and regulation-aligned ontology model that addresses critical challenges in safety information integration, sharing, and application.

1. Introduction

Construction remains one of the most hazardous industries, with a high incidence of workplace accidents reported annually. According to the Occupational Safety and Health Administration (OSHA), there are four primary types of fatal construction incidents—commonly referred to as the “Fatal Four”. In 2018, 1008 out of the 4779 fatalities in the private sector occurred in construction, representing 21.1% of the total. Falls were identified as the leading cause of these deaths, aside from highway-related incidents, which also significantly impact the sector. Data from the U.S. Bureau of Labor Statistics further indicates that the Fatal Four were responsible for 58.6% of all construction worker fatalities in 2018. Eliminating these hazards could potentially save the lives of approximately 591 workers in the U.S. each year (see

Table 1. Top four hazards in construction.

Fatal Four	Types of Causes	Deaths	%
1	Falls	338	33.5
2	Struck by object	112	11.1
3	Electrocutions	86	8.5
4	Caught in/between	55	5.5
	Total deaths in construction in 2018	1008	58.6

).

Despite regulatory efforts, the total number of construction fatalities has shown a concerning rebound in recent years, reaching 1075 in 2023—the highest since 2011. Fall-related deaths alone accounted for over 38% of these cases, underscoring their persistent dominance among job site hazards [1]. In addition to falls, increased fatalities in trenching and excavation activities, as well as growing concerns over long-term occupational illnesses such as silicosis, have prompted regulatory bodies like OSHA to expand the traditional “Fatal Four” scope [2]. These developments reflect both evolving and persistent safety threats across modern construction environments.

Among these, fall-related accidents are the most frequent and are largely preventable through proactive safety planning and adherence to established protocols. Recent studies confirm that falls from height remain the most serious and persistent safety issue in the construction industry, both in terms of frequency and severity. For example, OSHA reports indicate that nearly 40% of construction worker fatalities in 2023 were caused by falls to a lower level, a proportion that has remained largely unchanged over the past decade [3]. A 20-year analysis found a 14.9% increase in fall accidents in U.S. construction between 2008 and 2020, signaling a concerning trend despite regulatory efforts [4]. The criticality of fall-related risks stems not only from their prevalence but also from their unpredictability and the difficulty in enforcing consistent fall protection practices. Many fall incidents result from transient human and environmental factors, such as the improper use of personal protective equipment (PPE), inadequate guardrails, or unanticipated task changes, all of which are difficult to fully mitigate through traditional safety planning [4,5]. Furthermore, fall hazards are particularly complex in roof-related work due to varying elevations, temporary structural elements, and limited anchorage options. These characteristics make fall-from-height accidents both frequent and hard to control, underscoring the need for advanced detection and prevention mechanisms. In response, the recent literature has proposed sensor-based wearables, computer vision, AI prediction models, and ontology-based systems as promising methods for proactive fall risk detection and hazard recognition [1,6,7]. This study builds on these insights by targeting the modeling of fall-related safety knowledge through a domain ontology framework that specifically addresses roofing activities—an area identified as high-risk and under-supported in existing systems.

While OSHA has instituted comprehensive safety regulations for the construction industry, a significant gap remains in the effective dissemination and practical application of safety knowledge on site. Many workplace hazards persist due to the inadequate sharing of safety information and a general lack of awareness among workers. This underscores the critical importance of developing a structured and accessible system for sharing construction safety knowledge.

To facilitate systematic knowledge sharing, computer-aided frameworks are essential [8,9]. Ontology provides a foundational approach for developing such frameworks, as it enables the formalized representation and exchange of knowledge among intelligent systems [10]. In recent years, ontology-based methodologies have been increasingly adopted in the construction sector [11]. Ontologies in this context refer to structured frameworks that define key concepts, relationships, and rules within a specific domain—in this case, construction safety. They allow computers and systems to reason about hazards, regulations, and workflows in a consistent, reusable manner. However, their impact has often been limited due to insufficient domain specificity, a lack of standardization, and minimal integration with practical safety regulations [12]. These limitations are further illustrated by several recurring issues found in earlier ontology-based systems. Earlier systems were often one-off or narrowly scoped, limiting their generalizability across diverse construction contexts [13]. Furthermore, poor integration with BIM workflows and digital tools hampered their practical adoption, especially for field personnel with limited technical expertise [14]. Without user-friendly interfaces and mechanisms for continuous knowledge updates, these ontology-based safety systems risk becoming static and underutilized [15].

In addition, recent studies have shown that many ontology systems fail in real-world construction settings because they are either too abstract or not aligned with practical workflows. Limitations include weak integration with BIM tools, low usability for field personnel, and the high effort required for their ongoing maintenance. Without intuitive interfaces and dynamic update capabilities, ontology systems often remain underutilized in active project environments [16]. Ongoing efforts in developing construction-specific ontology databases are therefore crucial to overcoming these limitations [17].

As domain ontology modeling forms the backbone of ontology-driven knowledge systems, it plays a pivotal role in enabling accurate, consistent, and reusable knowledge representation in the construction domain [10]. Any ontology-based knowledge-sharing initiative must begin with the development of a robust domain ontology model [18].

The present study aims to integrate construction safety knowledge and Building Information Modeling (BIM) data through ontology-based modeling, thereby establishing a comprehensive and semantically structured knowledge base (see Figure 1). Figure 1 illustrates the overall research framework of this study, outlining the integration process between construction safety knowledge, domain ontology modeling, and BIM data. It shows how the system architecture enables a feedback loop between knowledge acquisition, ontology development, and BIM-based hazard representation, ultimately leading to improved decision-making support. Specifically, this research focuses on modeling fall-related hazards on roofs, which OSHA identifies as among the most common and fatal types of accidents in construction [19].

The ontology model developed in this study was created using Protégé, a widely recognized ontology editing platform. The model is structured into three main superclasses: the Construction Class, Conduct Class, and Safety Control Class. The Construction Class encapsulates detailed information on building components, derived from BIM platform such as Revit. The Conduct Class maps specific construction tasks, processes, and their associated hazards, particularly in roofing operations. The Safety Control Class incorporates safety incidents, rules, and countermeasures aligned with OSHA guidelines, focusing on roof-related safety standards.

This study presents an ontology-based construction safety knowledge system designed to enable computers to recognize, understand, and manage safety information. The proposed domain ontology facilitates the identification and analysis of potential risks and regulatory violations, supporting both safety planning and strategic decision-making. In particular, the novelty of this study lies in the development of a domain-specific ontology that semantically models fall-related hazards in roofing tasks—an especially high-risk area that has been underrepresented in prior research. Unlike many existing ontologies, which are either generic or overly abstract, the proposed model encodes OSHA regulations into machine-readable SWRL rules and integrates them with BIM-based building elements and construction task processes. This integration allows for automated reasoning and the real-time evaluation of safety compliance within practical project settings, offering a reusable and scalable framework for future ontology development in the construction safety domain.

2. Review

2.1. Related Works

Extensive research has been conducted to address interoperability challenges across diverse methodologies and application domains. As depicted in Figure 2, ontologies have been increasingly utilized as strategic tools to facilitate platform-independent knowledge sharing and semantic integration. Numerous studies have focused on developing ontology-based systems to enhance knowledge exchange within information technology platforms [20,21]. Aziz et al. [22] introduced a prototype application employing a semantic resource description framework to enable context-aware information delivery. In a related effort, Wang et al. [23] proposed a novel approach for managing situation-sensitive construction data. The usefulness of ontology in supporting classification systems for effective knowledge sharing was also emphasized in Wang et al.’s earlier study [18]. Several researchers have contributed to ontology-driven taxonomy construction and classification frameworks, including Rezgui [24], Elghamrawy and Boukamp [25], El-Diraby and Zhang [26], and Niu and Issa [27]. Lin et al. [20] presented an ontology-based extraction concept to ensure the relevance, accuracy, and completeness of domain knowledge, while Yurchyshyna et al. [28] proposed an ontological approach for verifying conformance in data systems. Benevolenskiy [29] advanced a process- and ontology-based framework to standardize simulation workflows in construction. Ontology frameworks have also been explored in sensor-based building monitoring systems [30], as well as in text-based knowledge processing. Zhou and El-Gohary [31] developed an ontology-based semantic classification model and later proposed a technique for automated information extraction.

Although these contributions have led to the development of numerous frameworks, systems, and applications, their practical implementation has often been constrained by the inherent complexity and variability of construction projects. Many systems require further refinement, empirical testing, and validation to reach their full potential.

Ontology has also gained traction in the design phase, particularly in conjunction with Building Information Modeling (BIM), due to its rich and structured representation of design information [32]. However, extracting complete and context-specific information from BIM models remains a challenge. To address this, research has highlighted the need for user-centric interfaces and expanded functionalities [24].

Kim et al. [33] explored inference, expression, and consistency checking at the design stage and later developed a tool to detect conflicts within BIM-based processes. Nepal et al. [32] proposed a feature-based querying system to extract construction-specific design conditions, enabling rapid and customizable data retrieval. Zhang and Issa [34] worked on refining BIM data extraction through IFC format enhancements, while Venegopal et al. [35] extended IFC capabilities using model view definitions. In a related line of work, Niknam and Kashkashenas [36] demonstrated a semantic architectural design knowledge base built through case studies.

Despite advancements in information management technologies, the application of ontology for managing construction safety knowledge remains limited [37]. Previous studies have largely focused on safety design, education, and policy development. There is now a pressing need to shift focus toward the utilization and sharing of safety knowledge and data to prevent construction-related accidents [38].

To this end, several ontology-based frameworks have been proposed for safety knowledge sharing and hazard analysis. Wang and Boukamp [23,39,40] developed ontology-based approaches for risk analysis, including a context-sensitive reasoning framework. Lu et al. [41] introduced an ontology for construction safety inspection, integrating emerging technologies into safety management systems. Zhang et al. [37] combined BIM and GIS to create an automated workspace visualization system for analyzing potential collisions among workers and machinery. However, despite these efforts, knowledge-based research for improving safety-related information exchange must continue to evolve [37]. Domain ontology development plays a foundational role in resolving interoperability and establishing comprehensive knowledge bases. Accordingly, continuous efforts are needed to expand, refine, and standardize domain ontologies across the construction safety domain [42].

2.2. Current Status Quo

This study undertook a review of the recent literature to assess prevailing themes and research directions in ontology applications in construction. The increasing volume of studies in recent years reflects a growing academic and practical interest in ontology research; between 2015 and 2024, the number of ontology-related publications in construction increased from 23 to approximately 45 (with a dip to 16 in 2017), as illustrated by the yearly publication trend graph inserted in the main text. This trend, which illustrates a growing interest in ontology due to digital transformation in construction, is expected to continue as construction projects become more data-intensive and reliant on information and communication technologies (Figure 3) [37].

The increasing complexity and size of construction projects demand robust systems for managing vast quantities of data, knowledge, and information. Ontology has emerged as a promising method for organizing this data semantically, thereby improving interoperability and enabling intelligent decision support. It is imperative that ongoing research continues to explore ontology’s application in the construction industry to meet evolving knowledge management needs.

In parallel, the importance of construction safety knowledge sharing is gaining recognition, especially as awareness of safety’s role in project success increases. However, research dedicated to the structured sharing of safety knowledge remains limited. Despite advancements in knowledge and data management technologies, the integration of these innovations into safety knowledge systems has not kept pace [37,43].

Earlier studies in the safety domain have primarily focused on accident causation, the effectiveness of safety designs, education programs, and policy interventions. In contrast, relatively few studies have examined how construction participants can effectively share safety-related knowledge and information [38]. Although several ontology-based frameworks have been introduced for work risk analysis and safety knowledge exchange, many of these remain focused on surface-level concepts. The absence of comprehensive domain ontologies continues to hinder the validation and reuse of these systems.

Thus, the sustained development of domain-specific ontologies is necessary, with particular emphasis on reflecting diverse safety regulations and requirements. Furthermore, proposed frameworks must be subjected to rigorous verification and performance evaluation before they can be effectively reused or scaled [37].

2.3. Challenges for Building Ontologies in the Construction Domain

Despite considerable progress, several critical challenges remain in ontology development for the construction industry. Notably, many studies have emphasized the foundational role of ontology in overcoming interoperability and structuring domain knowledge systems. However, several barriers must be addressed to realize its full potential.

First, there is a need to develop more domain-specific ontologies to enable robust and reusable knowledge bases within the construction field. Second, data heterogeneity must be addressed through the standardization of formats and definitions under well-defined conditions. Third, in terms of semantic structure, classes with ambiguous or multiple meanings should be disambiguated to improve the precision and utility of ontologies.

Fourth, existing ontology models require thorough validation to ensure their reliability and practical applicability. Lastly, the open sharing of developed ontology models is essential. While many projects independently develop ontological frameworks, these efforts are often siloed and rarely shared, resulting in duplicated efforts and missed opportunities for collaborative advancement. This lack of model reuse contradicts one of the core strengths of ontological modeling and hinders the development of standardized methodologies across the field.

Redundant modeling efforts not only delay progress but also contribute to inconsistencies in ontology structure, thereby obstructing the establishment of a unified framework for construction knowledge management.

3. Methodology

This study proposes a construction safety domain ontology, developed using the Protégé ontology editor, to define the structure of classes, relationships, and axioms. The consistency and coherence of the constructed ontology were evaluated using OWL reasoners and competency questions. To formalize safety knowledge based on OSHA regulations, Semantic Web Rule Language (SWRL) rules were also developed and implemented within the ontology framework [37]. The rule sets were designed to be adaptable, allowing domain experts to modify and extend them according to specific contextual requirements. Figure 4 outlines the generalized methodology for domain ontology development, which served as the foundation for this research.

3.1. Selecting Domain

Domain ontology serves as the foundational layer for constructing knowledge-based systems that enable interoperability in information technologies. The standardization of domain ontologies is a key enabler of interoperable and semantically coherent data exchange systems [43]. A domain ontology provides a shared conceptual framework for representing knowledge specific to a particular area [7]. As reviewed in Section 2, there is a growing need for structured research that facilitates the sharing of construction safety information. Despite its importance, construction safety remains an underexplored domain in ontology-based research. Accordingly, this study targets construction safety as the primary domain of interest.

3.2. Defining Work Scope

According to OSHA data, fall-related incidents constitute the highest proportion of accidents in the construction industry. Roof-related activities, in particular, present the most frequent and severe fall hazards. OSHA has identified that falls from roofs are among the most common fatal events in construction settings, with 888 events. Therefore, the scope of this study was specifically defined to encompass roof-related work processes. As shown in

Table 2. Roof-related construction work.

Task	Activity	Job Step
Roof	Framing	Premade trusses
		Mount trusses
		Install temporary braces
		Install end trusses
		Install standard trusses
		Install permanent bracing
	Sheathing	Provide layer of material
	Underlayment	Install underlayment
	Flashing	Install metal drip edge
	Singles	Apply singles

and

Table 3. Hazards of roof work.

Task			Roof
Activity			Framing	Sheathing	Underlayment	Flashing	Singles
Hazards	1	Roof Stability	x
	2	Ladder Security and Placement	x
	3	Weather Conditions		x	x	x	x
	4	Roof Holes		x	x	x	x
	5	Edge Awareness		x	x	x	x
	6	Improper Training	x	x	x	x	x
	7	Improper Use of Fall Protection Equipment	x	x	x	x	x
	8	Pitch		x	x	x	x
	9	Parapet Walls
	10	Loose Debris		x	x	x	x
	11	Extreme Heat		x	x	x	x
	12	Slippery Conditions		x	x	x	x
	13	Chemical Exposure		x	x	x	x
	14	Repetitive Motion Injuries		x	x	x	x

, the study comprehensively investigated the various tasks associated with roofing and identified potential hazards at each step. Further work requirements—based on the severity and frequency of safety risks—will be incorporated in subsequent studies.

3.3. Domain Ontology Development and Validation

A construction safety domain ontology was developed and validated as illustrated in Figure 5, which was revised to explicitly include the integration of SWRL rules and the validation step using the Pellet reasoner.

i.

Identification of Core Classes: Fundamental building components and their respective functions and purposes were first delineated.

ii.

Taxonomy Creation: Hierarchical structures for classes, their functions, and their use cases were defined.

iii.

New Concept Integration: Definitions for newly identified building elements were incorporated into the ontology.

iv.

Axiom Design: Logical expressions and rules were created to explain the relationships and constraints among the subclasses.

v.

Competency Questioning: A series of competency questions were formulated to validate the ontology’s reasoning capabilities.

vi.

SWRL Rule Integration: Semantic Web Rule Language (SWRL) rules were developed and embedded into the ontology to represent OSHA safety regulations in a machine-readable format, enabling rule-based reasoning within the Protégé environment.

vii.

Validation: The ontology model was verified using the Pellet reasoner to ensure logical consistency across classes, axioms, and object properties.

For the Safety Control Class, a structured process for work risk analysis was applied. First, all roofing-related tasks and associated work steps were cataloged. Second, potential hazards were identified for each task. Third, appropriate risk mitigation actions were linked to each task phase.

Ontology construction was carried out using Protégé. The vocabulary of terms and classes was partially sourced from Autodesk Revit and its associated library, given its widespread use in BIM-based modeling. While efforts were made to reuse existing ontological resources—such as environmental ontologies—many were deemed unsuitable due to their domain misalignment and outdated structures. As such, this study opted to develop custom ontology classes that better reflected the construction safety domain.

In addition to class hierarchy modeling, SWRL rules were developed and embedded within the ontology using Protégé to formalize OSHA safety regulations and support rule-based reasoning. All term definitions were referenced and refined using Termium Plus, Merriam-Webster, Wikipedia, and selected technical handbooks, and modified according to established ontology engineering principles [44]. To ensure semantic consistency and domain relevance, terms sourced from Revit (representing BIM component naming), OSHA (regulatory terminology), and Termium Plus (standardized language definitions) were systematically cross-validated. The cross-validation process involved mapping similar terms across the three sources, resolving naming inconsistencies, and selecting terminology that aligned with both construction site practices and regulatory contexts. In cases of conflict or ambiguity, OSHA definitions were prioritized for regulatory accuracy, while Revit terms were used to ensure compatibility with BIM environments. The finalized terms were then standardized and encoded into the ontology structure. The finalized ontology model was validated using reasoning tools and competency question queries. As shown in Figure 6 and Figure 7 the ontology was verified using Pellet, an integrated reasoner in Protégé. The consistency check confirmed that no logical inconsistencies existed within the class hierarchy, object properties, or axioms. Figure 5 illustrates the reasoning process, including the inferred class relationships and the successful result of the consistency validation in the Pellet output console.

3.4. Semantic Web Rule Development

According to OSHA standards, roofs are classified into two categories: low-slope and steep-slope roofs. A low-slope roof is defined under regulation 1926.500(b) as one with a slope less than or equal to 4:12 (horizontal to vertical). Conversely, a steep roof is defined in 1926.500(b)(2) as one with a slope greater than 4:12.

OSHA regulations 1926.501(b)(10) and 1926.501(b)(11) specify the mandatory use of safety systems for workers operating on low-slope roofs with unprotected edges exceeding six feet (1.8 m) in height. The required systems include guardrail systems, safety net systems, personal fall arrest systems, warning line systems, and safety monitoring systems. For work on steep roofs, OSHA mandates the use of guardrails with toe boards, safety nets, or personal fall protection equipment to mitigate fall risks.

In this study, the regulatory provisions outlined by OSHA were translated into machine-readable SWRL rules to enhance the semantic reasoning capabilities of the ontology. This transformation enabled the more precise integration of OSHA regulations with ontology-based risk detection and prevention mechanisms. As an illustration,

Table 4. Semantic Web Rule Language for OSHA requirements.

Scope	Standards	Contents
Roof	1926.500(b)	Low-slope roof means a roof having a slope less than or equal to 4 in 12 (vertical to horizontal).
	SWRL1	roof(?r)^has_Slope(?r,?s)^swrlb:lessThanOrEqual(?s, “18.43”^^xsd:float)-> low_slope_roof(?r)
	1926.500(b)(2)	Steep roof means a roof having a slope greater than 4 in 12 (vertical to horizontal).
	SWRL2	roof(?r)^hasSlope(?r,?s)^swrlb:greaterThan(?s, “18.43”^^xsd:float)-> steep_slope_roof(?r)
	1926.501(b)(10)	“Roofing work on Low-slope roofs.” Except as otherwise provided in paragraph (b) of this section, each employee engaged in roofing activities on low-slope roofs, with unprotected sides and edges 6 feet (1.8 m) or more above lower levels shall be protected from falling by guardrail systems, safety net systems, personal fall arrest systems, or a combination of warning line system and guardrail system, warning line system and safety net system, or warning line system and personal fall arrest system, or warning line system and safety monitoring system. Or, on roofs 50-feet (15.25 m) or less in width, the use of a safety monitoring system alone [i.e., without the warning line system] is permitted.
	SWRL3	low_sloped_roof(?r)^hasHeight(?r,?h)^swrlb:greaterThanOrEqual(?h, “6.0”^^xsd:float)->requires(?r, guardrail)^ requires(?r, safety_net_system)^ requires(?r,personel_fall_arrest_system)^requires(?r,warning_line_system)^ requires(?r, safety_monitoring_system)
	1926.501(b)(11)	“Steep roofs.” Each employee on a steep roof with unprotected sides and edges 6 feet (1.8 m) or more above lower levels shall be protected from falling by guardrail systems with toe boards, safety net systems, or personal fall arrest systems. Where the slope of the roof exceeds 35 degrees. The roof is an inappropriate surface to stand on. Perimeter guardrails and catch platforms are inappropriate measures to protect workers on a steeply sloping roof.
	SWRL4	steep_sloped_roof(?r)^hasHeight(?r,?h)^swrlb:greaterThanOrEqual(?h, “6.0”^^xsd:float)->requires (?r, guardrail)^ requires(?r, toeboard) requires(?r, safety_net_system)^ requires(?r, personel_fall_arrest_system)

presents the mapping of OSHA safety requirements into their corresponding SWRL rule representations, including SWRL3 and SWRL4, which address roof-related fall hazard scenarios. It should be clarified that the safety systems represented in SWRL3 and SWRL4 are not intended to be applied concurrently. Rather, they reflect a set of conditional alternatives as defined in OSHA regulations, contingent upon specific contextual factors such as roof slope and width. Accordingly, the ontology interprets these rule elements using conditional logic to preserve semantic alignment with regulatory intent and practical applicability in real-world scenarios.

4. Result

4.1. Three Core Superclasses and Semantic Web Rule Language

This study aims to formalize construction safety knowledge through the development of a construction safety domain ontology. The proposed ontology serves as a foundational knowledge base that supports safety analysis and management by semantically linking safety knowledge to both construction processes and physical components. Moreover, it establishes a critical infrastructure for the development of automated systems capable of reasoning about safety in construction environments.

The ontology is structured around three primary superclasses: the Construction Class, Conduct Class, and Safety Control Class (see Figure 8). In addition to this conceptual framework, the study introduces Semantic Web Rule Language (SWRL) rules aligned with OSHA regulations to enable logical inference within the ontology (see Figure 9). SWRL plays a key role in connecting classes and their properties based on regulatory logic, thus enhancing the model’s capability to simulate real-world safety compliance scenarios.

• Construction Class: This superclass encompasses building components commonly modeled in Building Information Modeling (BIM) environments, including columns, slabs, walls, foundations, beams, frames, and roofs. These subclasses establish the structural basis required to connect the ontology with BIM systems, enabling the integration of ontology-based safety reasoning into digital models.
• Conduct Class: This class represents construction activities and processes, including labor, equipment, and material usage. It contextualizes tasks and their associated risks within various phases of the construction project.
• Safety Control Class: This superclass comprises hazard classifications, safety regulations, mitigation strategies, and safety equipment requirements. It represents domain-specific safety knowledge, facilitating the analysis of compliance and risk mitigation strategies.

Through these interlinked classes and the integration of SWRL rules based on OSHA standards, the proposed ontology enables semantic linkage between BIM models and safety knowledge. This facilitates automated compliance checking and risk identification in the early stages of construction planning and design.

4.2. Discussion

The ontology developed in this study provides a comprehensive knowledge framework for representing construction safety information and was designed to operate seamlessly with BIM environments. By incorporating OSHA regulations into SWRL rules, the ontology supports automated reasoning that can be applied during the design phase to evaluate safety compliance and predict potential hazards. The modeling process was carried out using the Protégé ontology editor, with careful attention paid to the definitions of classes, relationships, and axioms. The logical consistency of the domain model was validated using the built-in OWL Reasoner. OSHA safety regulations were translated into SWRL rules to ensure machine-readable rule enforcement aligned with regulatory guidelines [38]. One of the intended benefits of ontology modeling is the reuse of existing semantic structures. Initially, this study sought to adapt classes from existing ontologies, particularly those developed in the biomedical domain, where ontology-based modeling is more mature. However, most available ontologies lacked domain specificity for construction applications. Moreover, the classes identified were often outdated or defined without construction context, requiring significant modification. Consequently, the study determined that constructing new, domain-specific classes would be more efficient and contextually appropriate than attempting to reuse and revise unrelated models. This decision was also influenced by the lack of accessible, shared ontology repositories for the construction safety domain, highlighting a gap in semantic infrastructure. Additionally, standardizing safety knowledge proved challenging due to the inherently diverse formats, terminologies, and contextual interpretations used across the construction industry, which often lack semantic consistency and interoperability. Term definitions were refined using references such as Termium Plus, Merriam-Webster, Wikipedia, and construction-related handbooks, and adjusted to meet ontology design principles [37,38]. The overarching contribution of this research lies in its creation of a machine-readable, semantically structured safety knowledge base. This ontology allows for the clear and logical encoding of safety concepts that can be directly used in future applications, including intelligent safety analysis tools and real-time decision-support systems. To ensure the reliability of the ontology’s outputs, two validation approaches were employed. First, logical consistency was verified using the Pellet reasoner in Protégé to confirm that the ontology was free of contradictions across its classes, axioms, and relationships. Second, a set of competency questions was used to validate whether the ontology could correctly infer safety compliance conditions based on encoded OSHA regulations. These methods demonstrated the model’s ability to produce accurate and context-relevant outputs. To transform these results into real-world applications, the proposed ontology can be integrated into existing BIM-based design and safety management workflows. Specifically, the ontology can serve as a back-end reasoning engine within BIM tools (e.g., Autodesk Revit, Navisworks) by linking BIM elements with corresponding ontology classes and SWRL rules. This would enable automated safety compliance checking during the design phase, where high-risk elements such as unprotected roof edges can be identified and matched with OSHA-mandated protective measures. Moreover, the ontology framework can be implemented in safety inspection software or digital construction management systems used by contractors or public clients to support risk analysis, reporting, and training. Such integration bridges the gap between semantic knowledge representation and operational decision-making, allowing the ontology to be applied in real construction planning, supervision, and review processes.

In addition to these conceptual integrations, this study envisions the ontology being deployed directly in real-world workflows through its linkage with BIM-based project platforms and field-level safety operations. By embedding the ontology engine in safety review modules during pre-construction planning, project teams can automatically identify violations of OSHA safety standards before site work begins. Furthermore, during the construction phase, real-time BIM dashboards enhanced with ontology reasoning can be used by site managers to identify high-risk areas, prompt recommended interventions, and support compliance monitoring. The system’s ability to generate rule-based safety evaluations also makes it a practical tool for use in regulatory inspections or audits, particularly for public-sector projects. Through these mechanisms, the ontology is not only a theoretical knowledge structure but a directly applicable asset for real-time safety management and policy enforcement in construction practice.

In future work, we plan to conduct empirical validations using real-world BIM data and expert evaluation to further reinforce the practical utility of the proposed knowledge model. Looking ahead, this study aims to contribute to the establishment of a Construction Safety Ontology Library, serving as a central resource for domain-specific ontologies. This library will support knowledge sharing, standardization, and continuous development within the construction safety domain. Such infrastructure is essential to advancing digital transformation and safety intelligence in the architecture, engineering, and construction (AEC) industry.

5. Conclusions

While ontology-based knowledge management has been extensively explored in the construction domain, its application to construction safety remains relatively underdeveloped. As domain ontologies form the foundational structure for knowledge systems that enable efficient information utilization, the development of a domain-specific ontology for construction safety is both necessary and timely. This study addressed this gap by focusing on the safety domain—particularly fall hazards associated with roofing tasks—as its primary area of investigation.

To that end, the study identified relevant risks and constructed an ontology domain model comprising three primary superclasses. The Construction Class encapsulates core building elements based on BIM frameworks. The Conduct Class represents the sequence of construction activities and tasks, detailing the materials, equipment, and labor associated with each process. Lastly, the Safety Control Class integrates safety-related information, including regulatory requirements, best practices, and procedural guidance for mitigating risk during specific construction operations. To enhance the ontology’s utility in intelligent systems, semantic rules were encoded using the Semantic Web Rule Language (SWRL). These rules translate OSHA safety guidelines into a machine-readable format, allowing automated reasoning systems to infer potential hazards, predict accident types, and recommend appropriate countermeasures in scenarios where safety compliance is breached. Furthermore, the study not only proposed but also validated a domain ontology tailored to the construction safety domain, while addressing potential modeling challenges and practical implementation barriers. The resulting ontology supports the development of automated safety analysis systems integrated within BIM environments. As ontology reuse significantly reduces the time and effort required by domain experts, the proposed model can serve as a foundational framework for future ontology extensions in safety-related applications. It offers a reusable structure for constructing complex relationships between safety entities, thereby promoting scalability and efficiency in domain-specific modeling.

Despite these contributions, several limitations should be acknowledged. First, the ontology model was validated primarily through logical consistency checking and competency questions, without full-scale deployment or empirical testing in active construction projects. Second, the scope of the ontology was limited to fall-related hazards in roofing tasks, which, while significant, does not encompass the broader range of safety risks encountered in construction. Third, integration with BIM platforms and external safety systems was conceptually discussed but not technically implemented or evaluated within this study. Based on these limitations, future research should focus on three key directions: (1) conducting empirical validation through real-world pilot applications in BIM environments; (2) expanding the ontology to include additional hazard categories such as electrical, struck-by, or caught-in-between incidents; (3) developing plug-in tools or APIs for the seamless integration of the ontology with commercial BIM software to enable automated compliance checking and real-time safety feedback during design and construction.

Beyond its technical contributions, this study also offers important implications for construction safety policy development. By formalizing OSHA regulations into machine-readable SWRL rules and integrating them within a structured ontology framework, the proposed model provides a transparent, traceable, and logic-driven foundation for evaluating safety compliance. This can aid regulatory agencies and policymakers in identifying regulatory gaps, benchmarking current safety standards, and developing data-supported safety policies. Furthermore, the ontology’s structured representation of risk scenarios and mitigation measures enables the more consistent interpretation and application of safety regulations across diverse construction projects, promoting greater regulatory alignment and enforcement. The ability of the ontology to simulate rule-based compliance checks within BIM environments also supports evidence-based policymaking by allowing policymakers to evaluate the effectiveness of safety regulations and fall-prevention strategies under realistic construction conditions before implementation. In doing so, the system serves not only as a knowledge model but also as a practical simulation and decision-support tool, helping bridge the gap between policy design and on-site execution.

In conclusion, this research successfully fulfills its original objective stated in the introduction: to model fall-related safety knowledge in roof work using a domain-specific ontology and to integrate that knowledge with BIM-based data structures to support automated safety reasoning. Specifically, the proposed ontology enables several practical functions: (1) the automated identification of safety hazards based on OSHA-compliant rules; (2) semantic linking between construction tasks, physical BIM elements, and associated risks; (3) the machine-readable representation of safety regulations to support real-time compliance checking and intelligent decision-making. These capabilities provide the groundwork for intelligent safety analysis tools that can be embedded within BIM environments or digital construction platforms. Through the achievement of these goals, the study contributes meaningful progress in both semantic safety modeling and its practical application within construction informatics. Looking ahead, future research should aim to address current limitations, broaden the model’s domain scope, and implement the ontology within operational construction environments to maximize its real-world impact.