CargO-S: A pattern-based well-founded legal domain ontology for the traceability of goods in logistic sea corridors

Abstract

Building legal domain ontologies is a prominent challenge in the ontology engineering community. The ontology builders confront issues such as the complexity of the legal domain, the difficulty of applying existing ontology engineering approaches, and the intention of developing legal models faithful to realities. In this paper, we discuss constructing a well-founded legal domain ontology, named CargO-S, for the traceability of goods in logistic sea corridors. For building CargO-S, a pattern-oriented approach is applied, supported by ontology-driven conceptual modeling, ontology layering, and ontology reuse processes. CargO-S is grounded in the unified foundational ontology UFO by using the ontology-driven conceptual modeling language OntoUML. Besides, ontology layering is proposed to simplify the development process by dividing CargO-S into three layers located at different granularity levels: upper, core, and domain. For building the upper and core layers, conceptual ontology patterns are reused from the foundational ontology UFO and the legal core ontology UFO-L. These patterns are applied, either by extension or analogy with legal rules, for building the domain layer. CargO-S is then validated by implementing the ontology as OWL and SWRL rules. Finally, the performance and the semantic accuracy of CargO-S are evaluated using a dual evaluation approach.

Keywords

Legal ontologies well-founded ontologies ontology-driven conceptual modeling ontology patterns ontology reuse UFO UFO-L maritime law

1. Introduction and motivation

Legal ontologies are generalized conceptual models of specific parts of the legal domain (Mommers, 2003). They provide stable foundations for knowledge representation in this domain (Mommers, 2003). They have been developed and used for legal knowledge management and as knowledge bases in legal knowledge systems (Benjamins et al., 2005). Legal ontologies have been useful in several applications to support semantic indexing, information retrieval, semantic integration, reasoning, and problem-solving (Breuker et al., 2004; Wyner, 2009; Ashley, 2009; Wyner and Hoekstra, 2010). Legal domain ontologies differ from ontologies in other fields of practice, like medicine or engineering. For Breuker et al. (2004), the law is aimed at social activities, which are generally described in common sense terms. Thus, legal ontologies have to cover a wide range of common-sense concepts that are part of physical, abstract, mental, and social worlds (Benjamins et al., 2005). Besides social activities, legal domains share complex and varied notions of norm and responsibility representing the legal core notions (Benjamins et al., 2005). Examples of legal core notions are norm, contract, agent, role, normative position (e.g., duties, rights, etc.), responsibility, provision, sanction, and legal document. The combination of world and normative knowledge makes up the resource for reasoning in legal domains (Breuker et al., 2004). In the ontology engineering domain, the legal knowledge complexity causes some difficulties for building “factual” legal domain ontologies. Legal conceptual knowledge is closely related to the use of language (utilized in legal documents) that is too complicated (Francesconi and Tiscornia, 2008). Legal rules and standards are written, for the most part, in ordinary language containing ambiguities (Dove, 1996). Specifically, we cite the incomplete definition of many legal concepts of the law that Gardner (1987) originally addressed. Therefore, extracting semantic knowledge from juridical textual resources such as regulations and codes is a time-consuming and challenging task.

In CLASSE21

¹
http://classe.projet-recherche-normandie.fr/index.php, last visited January 18, 2021.

project, the main goal is to study the flow of goods in the logistic corridors of Upper Normandy in France in the context of the international maritime law. Our further objective is to build an automated application for the traceability of goods in these corridors. Implementing such an application implies creating an appropriate legal framework (Daduna et al., 2012). Such a framework requires an ontology-based model that reflects as much as possible the real application domain represented by the Hague-Visby Rules (Hague-Visby, 1968) on carriage of goods by sea. The ontology model, which will be employed as a knowledge base for data querying and decision support purposes, should represent “faithfully” the main legal agents, legal objects, legal events, and legal relationships of the domain. The ontology models that are faithful to realities are called well-founded domain ontologies. This concept has been used mainly in Guizzardi’s works (Guizzardi, 2005), and it refers to ontologies that are grounded in validated foundational ontologies. In other words, concepts and relations in a well-founded domain ontology are previously analyzed in the light of a foundational ontology. An example of a useful situation of the targeted ontology is the semantic search of the events of carriage of goods by sea, the involved legal agents, the accidents of loss or damage of goods, and the situations triggering such loss. Besides, the ontology supports the decision of liabilities in case of loss or damage of goods. For instance, based on the violated legal rule, which is considered in the ontology, the liabilities of the involved legal agents are defined.

To achieve our goal, reusing foundational and/or core ontologies is recognized as a promising approach (Falbo et al., 2013a). Foundational ontologies such as UFO (Guizzardi, 2005), and DOLCE (Borgo and Masolo, 2010), define a range of top-level domain-independent ontological categories that form a general foundation for more elaborated domain-specific ontologies. Core ontologies such as UFO-L (Griffo, 2018), and LKIF-CORE (Hoekstra et al., 2009) in the legal domain provide a precise definition of structural knowledge in a specific field that spans across different domain applications. Ontology reuse can also be accomplished using modeling solutions such as ontology patterns (Falbo, 2014). Ontology Patterns (OPs), which favor reuse of encoded experiences and good practices, can help to solve recurrent ontology development problems (Presutti et al., 2009). They describe particular recurring modeling problems that arise in specific ontology development contexts, and present well-proven solutions for the problems (Guizzardi, 2005; Falbo et al., 2013a). Conceptual OPs are small fragments of conceptual ontology models that address a specific modeling issue and can be directly reused by importing them into the ontology under development. These patterns are to be used during the ontology conceptual modeling activity. Conceptual OP can extract a fragment of either a foundational or a core reference ontology, which constitutes its background (Gangemi, 2005). Reference ontologies are a particular kind of conceptual model developed to make the best possible description of the domain in reality (Guizzardi, 2007). The implementation of these models as machine-readable ontologies are named operational ontologies (Falbo et al., 2013b). The reference ontologies provide the taxonomic and axiomatic context of the extracted patterns. Thus, the extracted conceptual ontology patterns inherit the axiomatization (and the related reasoning service) that existed in the reference ontologies (Gangemi, 2005). In the ontology engineering community, OPs have been addressed mainly in the works of Presutti et al. (2009); Gangemi and Presutti (2009); Blomqvist et al. (2009); Gangemi (2005). Recently, this approach has gained more attention, especially in Falbo et al. (2013a); Ruy et al. (2015, 2017), where its primary goal is to support the building of more consistent ontologies in a reuse-centered process.

The main contribution of this study is to develop and evaluate a well-founded legal domain ontology, named CargO-S, for representing the domain of carriage of goods by sea. For building CargO-S, a pattern-oriented approach is applied, supported by ontology-driven conceptual modeling, ontology layering, and ontology reuse processes. CargO-S is grounded in the unified foundational ontology UFO by applying the ontology-driven conceptual modeling language OntoUML. Besides, the structure of CargO-S is divided into three layers located at different granularity levels: upper, core, and domain. For building the upper and core layers, conceptual ontology patterns are reused from the foundational ontology UFO (Guizzardi, 2005) and the legal core ontology UFO-L (Griffo, 2018). These patterns are applied either by extension or analogy with the legal rules to build the domain layer. CargO-S is then validated by implementing the ontology model as OWL and SWRL rules. Finally, a dual evaluation approach is implemented to assess the performance and the semantic accuracy of CargO-S. The remainder of this article is organized as follows. Section 2 outlines UFO and UFO-L as this study’s background. The methodology of building CargO-S is presented in Section 3. Section 4 discusses the building process of CargO-S. In Section 5, the implementation of CargO-S is introduced. The evaluation of CargO-S is presented in Section 6. In Section 7, the related work is overviewed. Finally, Section 8 and Section 9 discuss and conclude this study respectively.

2. Background: UFO and UFO-L

In this section, we outline the background of this work: the Unified Foundational Ontology (UFO) and the legal core ontology UFO-L.

2.1. The unified foundational ontology (UFO)

The Unified Foundational Ontology (Guizzardi, 2005) is a reference foundational ontology that employs results from formal ontology, cognitive psychology, linguistics, and philosophical logic. UFO makes a fundamental distinction between Individuals and Universals. Individuals are entities that exist in reality and obey a unique and determinate principle of identity, while Universals are abstract patterns of features that can be realized in several different individuals (Guizzardi, 2005). In UFO, two main kinds of Individuals are distinguished: Endurants and Perdurants. Endurants are wholly present whenever they are present, i.e., they do not have temporal parts (Guizzardi et al., 2016). They can be further specialized into Substantials (Objects) and Moments (Tropes) (Guizzardi and Wagner, 2010). Substantials are existentially-independent Endurants (e.g., a ship, a person). Moments, or Tropes, in contrast, are individuals that can only exist by inhering in other individuals (Guizzardi and Wagner, 2010). Two main types of Moments are distinguished in UFO: Intrinsic Moments and Relators. Intrinsic moments are moments that inhere in one individual. An example of an intrinsic moment is a Mode (e.g., intention, duty). Relators are moments that depend on two or more Endurants (e.g., a business contract). Perdurants (Events) are individuals composed of temporal parts and are existentially dependent on Endurants (Guizzardi, 2013). Examples of Perdurants are a war, a strike, etc. Thereby, UFO is composed of three main layers: UFO-A (Guizzardi et al., 2008b) (ontology of endurants), UFO-B (Guizzardi, 2013) (ontology of events), and UFO-C (Guizzardi et al., 2008b) (ontology of social entities). In Fig. 1, a fragment of UFO-C is depicted. UFO has been employed in the design of the ontologically well-founded conceptual modeling language OntoUML. OntoUML is proposed by Guizzardi (Guizzardi, 2005) based on the need for an ontology-based language that would provide the necessary semantics to construct conceptual models using concepts faithful to reality (Benevides et al., 2009). OntoUML uses the ontological constraints of UFO as modeling primitives and is specified above the UML2.0 meta-model (Guizzardi, 2005). Examples of some OntoUML’s modeling primitives are: kind, subkind, mode, relator etc. Over the years, OntoUML has been adopted by many research, industrial and government institutions worldwide (Guizzardi et al., 2015). This language has been successfully employed in several industrial projects in several domains such as Petroleum and Gas, News Information Management, E-Government and Telecommunication (Guerson et al., 2014). To build, evaluate, and implement OntoUML models, a model-based environment is needed, such as OLED2

²
OntoUML Lightweight Editor, https://github.com/nemo-ufes/ontouml-lightweight-editor, last visited January 18, 2021.

(Guerson et al., 2015).

Fig. 1.

A fragment of UFO-C (adapted from Guizzardi et al., 2008b).

2.2. UFO-L

UFO-L (Griffo, 2018) is a legal core ontology that uses the domain-independent concepts, relations and properties existing in UFO to represent essential concepts of law based on Alexy’s theory of fundamental rights (Alexy, 2009). UFO-L defines a variety of basic legal core concepts, such as legal norm, legal agent, legal relator, legal moment, and legal normative description (Fig. 2). In UFO-L, legal roles, which are prescribed by legal norms, are assigned to agents and played within the scope of legal relations. These relations are represented based on Alexy’s theory and reified using legal relators, which are relational entities existentially dependent on legal roles (Griffo et al., 2016a,b, 2018). In UFO-L, different legal roles are defined, such as Right Holder and Duty Holder. A Right Holder is someone who has a right to something against a Duty Holder. A Duty Holder is someone who has the duty to materialize the right of a Right Holder (Griffo et al., 2016b). Moreover, UFO-L defines various legal moments such as, Right to an Action, Duty to Act, Right to an Omission, and Duty to Omit. Concerning the legal relators, they are composed of simple, such as Right-Duty Relator, NoRight-Permission Relator, and Power-Subjection Relator, and complex legal relators, such as Liberty Relator. Right-Duty Relator uses the legal relation right–duty (correlative) to bind right holder and duty holder (Griffo et al., 2016b).

Fig. 2.

Excerpt of UFO-L (adapted from Griffo et al., 2016b).

Fig. 3.

Right_Duty_to_an_Action_Relator pattern (adapted from Griffo, 2018).

Fig. 4.

Power_Subjection_Relator pattern (adapted from Griffo, 2018).

Fig. 5.

Disability_Immunity_Relator pattern (adapted from Griffo, 2018).

In UFO-L, various modeling patterns for the legal relators are defined (Griffo, 2018) (e.g., Right_Duty_to_an_Action_Relator (Fig. 3), Power_Subjection_Relator (Fig. 4), and Disability_Immunity_Relator (Fig. 5)). In UFO-L, the modeling patterns are defined based on a list of competency questions such as (Griffo, 2018):

Which legal roles are involved in the legal relationship?

What are the legal moments that make up the legal relationship?

Who are the holders of each legal moment?

On whom is each legal moment externally dependent?

What is the event that grounds the legal relationship?

Is there a legal rule that defines the legal relationship?

2.3. Why UFO and UFO-L?

In our work, we have selected UFO and UFO-L as foundations for ontology reuse and grounding processes. Concerning UFO, its is considered as a foundational ontology that comprises a rich theory of relations and complex relational properties that are absent in other foundational ontologies (Guizzardi, 2006; Guizzardi et al., 2015). UFO has been successfully applied in a large number of domains ranging from natural science domains such as Petroleum and Gas to social domains such as organizations, services, and software (Falbo et al., 2010). Besides, the availability of the conceptual modeling language OntoUML that is founded on this ontology which permits the building of ontologies by reusing the modeling primitives of UFO as generic concepts (e.g. kind, subkind, relator, role, role mixin) (Griffo et al., 2016b). The ontologist can represent the categories of a given domain using these primitives without the need to rebuild them (Griffo et al., 2016b).

UFO-L is a legal core ontology developed based on UFO to represent Alexy’s theory of fundamental rights (Alexy, 2009). Alexy’s theory, which classifies norms as rules and principles (Griffo et al., 2016b, 2018), is based on an analysis of constitutional rights as principles, which are considered fundamentally different from rules (Klatt, 2012). Norms are classified as deontological norms and axiological norms. Deontological norms are, in turn, classified as rules and principles (Griffo et al., 2016b). Alexy’s theory, in contrast to other approaches in the literature such as Kelsen’s theory (Kelsen, 2005), include modern concepts of Law introduced by the explicit countenance of a social reality (Griffo et al., 2016b, 2018). On the other side, Kelsen’s theory sees Law as a set of norms (Raz, 1974). Kelsen assumes that the existence of basic norms is necessary to explain the unity and normativity of legal systems. Norms exist only if authorized and entailed by other norms forming chains of validity. Besides the classification of norms, Alexy’s theory considers legal positions which are defined as situations in which a subject, in a legal relation, has a right against other subject (Griffo et al., 2016b). Legal positions divide rights in rights to something, liberties, and competences (Griffo et al., 2016b; Griffo, 2018; Griffo et al., 2018).

3. Methodology

Inspired by SABiO (Falbo, 2014), a methodology for building well-founded domain ontologies, we have defined six phases for building CargO-S: (1) ontology specification, (2) ontology requirements, (3) architectural design, (4) conceptualization, (5) implementation, and (6) evaluation. The development process of CargO-S is supported by ontology layering, ontology reuse, and ontology-driven conceptual modeling processes. This approach is performed under the supervision of the domain experts.3

³
https://www.idit.fr/, last visited January 18, 2021.

3.1. Ontology specification

The ontology specification designates the scope and purpose of the intended ontology. CargO-S aims to represent the international maritime law, specifically Hague-Visby (Hague-Visby, 1968) and Hamburg (Hamburg, 1978) conventions. This study concerns Hague-Visby Rules composed of legal norms that govern the rights, duties, liabilities, and responsibilities of shippers and carriers of marine cargo. For instance, Fig. 6 depicts Article I that defines the basic concepts of the carriage of goods by sea (Carrier, Contract of carriage, Goods, Ship, and Carriage of Goods) and Article IV that discusses the liability of the carrier and the shipper in case of loss or damage of goods.

Fig. 6.

Excerpt of Hague-Visby Rules.

Concerning the purpose, Cargo-S will be used as a knowledge base in a system intended for the traceability of goods in logistic sea corridors. Traceability is a procedure that is increasingly demanded in the logistics area.4

⁴

www.stocklogistic.com, last visited July 31, 2020.

It refers to the set of procedures and measures aimed at identifying a good from its origin to its final destination. Generally, traceability helps to know the condition of the goods, the treatments attributed to the articles, or the distribution process that has been followed. In our study, two main tasks are needed for the traceability of goods:

Querying data: requesting information related to the transportation of goods, such as the origin and destination of goods, the contracts established between the legal agents, and the events that occurred during transportation.

Decision support: specifying the rights, duties, liabilities, and responsibilities of the different legal agents involved in the transportation of goods. For instance, in case of loss or damage of goods while carriage, there is a need to detect the liable agent (e.g., carrier, shipper, etc.) who will be punishable by indemnities. In this context, generally, a claimant of damage or loss of goods could be the consignee, the shipper, or the carrier of the goods (Hague-Visby Rules (Hague-Visby, 1968), Hamburg Rules (Hamburg, 1978)). Thus, the relations of liabilities could be maintained using different scenarios depending on whether it is explicitly or implicitly defined in the legal rules. For example, Article III (5), defines the liability’s relation in case of inaccuracies explicitly using the verb “indemnify”. Thereby, a claim of the carrier, which is considered a right holder, is maintained against the shipper, which is seen as a duty holder. Besides, this legal rule also limits the responsibility and liability of the carrier under the contract of carriage to any person other than the shipper.

Article III. (5) […] the shipper shall indemnify the carrier against all loss, damages and expenses arising or resulting from inaccuracies […] The right of the carrier to such indemnity shall in no way limit his responsibility and liability under the contract of carriage to any person other than the shipper.

However, in legal rules such as Article IV (1), the claimant is not defined explicitly in case of unseaworthiness. In this case, we refer to Article I specifically (a) and (b) that defines the contract of carriage of goods, which is covered by a bill of lading, to regulate the relations between the carrier and the shipper. Thereby, any infringement or violation of obligations will expose a liability’s relation between these two agents.

Article IV. (1) Neither the carrier nor the ship shall be liable for loss or damage arising from unseaworthiness […]

Article I. (a) ‘Carrier’ […] who enters into a contract of carriage with a shipper […]

Article I. (b) […] bill of lading relates to the carriage of goods by sea […] such bill of lading regulates the relations between a carrier and a holder of the same.

The liability of the carrier and the shipper towards the consignee is out of this work’s scope because the latter is not defined explicitly in Hague-Visby Rules, neither his rights, duties, or responsibilities.

3.2. Ontology requirements

In this section, we discuss the main ontological requirements that CargO-S should satisfy. As in software engineering domain, ontology requirements are divided into functional and non-functional (Mylopoulos et al., 2007). The functional requirements, which can also be seen as content-specific requirements, define what needs to be expressed by the ontology model. Meanwhile, the non-functional requirements specify how an ontology needs to be designed in order to be applicable. Concerning the functional requirements, they are defined based on the Hague-Visby Rules. In the following, the main functional requirements of CargO-S are introduced:

FR1 – To represent the main agents and objects as well as their hierarchical structure: the ontology should represent the different kinds of agents (e.g., carrier, shipper) and objects (e.g., goods, vessel, legal document), and their mereological structure (e.g., carrier is a legal agent, vessel is a physical object).

FR2 – To describe the legal events and their mereological and causal structure: in the ontology, a description of legally defined events (e.g., carriage of goods, loss of goods) and their mereological and causal structure is required. The ontology model should distinguish between the complex events and the atomic events. Besides, the causal events, or situations, that cause other events should also be characterized (e.g., the fire on the vessel caused the damage of goods).

FR3 – To define the legal relationships and their hierarchical and mereological structure explicitly: in CargO-S, it is mandatory to represent the legal relationships such as contract of carriage of goods by sea, and their hierarchical and mereological structure (e.g., contract of carriage of goods by sea is a right-duty legal relationship).

FR4 – To state precisely the participation of legal agents and legal objects in the legal relationships: the ontology should be able to represent how agents and objects are involved in legal relationships (e.g., the participation of the carrier and the shipper in the contract of carriage of goods).

FR5 – To define the legal moments and their mereological structure: by establishing the legal relationships, such as the contract of carriage of goods, different legal moments are defined, such as right, duty, responsibility, etc. In the ontology model, these legal moments should be represented as well as their mereological structure.

FR6 – To designate the legal moments that make up the legal relationships in which the agents are involved: the ontology must describe the legal moments (e.g., right to carry and duty to carry) that compose the legal relationships (e.g., contract of carriage of goods) in which the legal agents have participated (e.g., carrier and shipper).

In the other hand, the non-functional requirements comprise four main aspects:

NFR1 – The modularity of the ontology structure to decrease the complexity of the building process and encourage its reusability.

NFR2 – The ontology model should distinguish between the reference and the operational versions of CargO-S.

NFR3 – The ontology model needs to provide a clear separation of the structural knowledge from the domain knowledge.

NFR4 – The ontology should be shareable and applicable for building (semi-) automated applications.

NFR5 – The extensibility of the ontology to include future aspects.

3.3. Architectural design

In this phase, we present the architectural design of CargO-S which aims to fulfill the non-functional requirement NFR1. To simplify the building process of CargO-S, we propose applying ontology layering that divides the ontology structure into three layers located at different granularity levels (Fig. 7): upper, core, and domain. The upper layer, which is located at the most high level, contains abstract, or domain-independent, categories (e.g., Event, Agent, Agent_Role, and Object). Underneath the upper layer, the core layer is situated. This layer contains the domain-dependent, e.g., legal domain, field-independent categories (e.g., Legal_Event, Legal_Agent, Legal_Role, and Legal_Object). At the lowest level, the domain layer is located. This layer is domain-dependent and field-dependent, e.g., maritime law. It defines legal categories related to the domain of carriage of goods by sea (e.g., Carrier, Shipper, Carriage_of_Goods, and Bill_of_Lading).

3.4. Conceptualization

The conceptualization phase permits the building of the reference model of CargO-S in order to fulfill the non-functional requirement NFR2. Thus, a pattern-oriented top-down approach is applied, supported by ontology-driven conceptual modeling and ontology reuse processes. The building process starts by building the upper and core layers independently. Furthermore, the domain layer is developed based on the upper and core layers’ content and structure.

The ontology-driven conceptual modeling (ODCM) process is described as the application of ontological analysis based on foundational ontologies to improve the theory and practice of conceptual modeling (Guizzardi et al., 2008a). In our work, ODCM is applied, aiming to ground the ontology model of CargO-S in the foundational ontology UFO (El Ghosh and Abdulrab, 2019). Thereby, the concepts and relations of each layer are analyzed and represented using the modeling primitives of UFO in OntoUML (e.g., $kind$ , $subkind$ , $relator$ , $role$ , $rolemixin$ , etc.). Thus, during the conceptualization process, the OntoUML constraints for relating these types are respected (see Fig. 7 for an example). Moreover, ontological patterns and modeling rules inherent to OntoUML are applied to achieve consistent conceptual models (for more details, refer to (Barcelos et al., 2013)).

Fig. 7.

The layered structure of CargO-S.

The ontology reuse process aims to simplify the building process by selecting and reusing conceptual ontology patterns (COPs) from UFO and UFO-L. To fulfill the non-functional requirement NFR3, we have defined two main perspectives to specify the intended patterns: content perspective and structure perspective.

For the content perspective: domain-independent ontology patterns (DIOP) are extracted from UFO for constructing the upper layer and domain-dependent ontology patterns (DDOP) are extracted from UFO-L for building the core layer.

For the structure perspective, two main types of ontology patterns are specified: recognition and template. The recognition pattern represents a recurring set of concepts and relations (Gangemi and Presutti, 2009). Meanwhile, the template pattern represents a common perspective on how to solve a specific problem (Gangemi, 2007). This pattern, which facilitates the modelling of legal norms, could be reused in many cases representing some notion of best practices to solve modeling problems (Gangemi, 2007).

In the following, the main activities of the conceptualization process are introduced:

Selection and reuse of conceptual ontology patterns: the derivation of the conceptual ontology patterns is performed for each layer independently. Thus, a fragmentation process is performed to extract sub-ontologies as ontology modules from UFO and UFO-L for building the upper and core layers, respectively. The selection process is guided mainly by a list of Competency Questions (CQs) to define the scope of patterns based on the functional requirements. The conceptual ontology patterns are represented and verified syntactically within OLED to assess if they are built correctly. The verification is performed automatically according to the formal constraints defined in UFO.

Composition of conceptual ontology patterns: the extracted conceptual ontology patterns need to be organized and consolidated in order to build the complete structure of each layer. For this purpose, a composition process is applied within OLED to relate the patterns within the same layer. Thereby, the consistency of the composition process is verified in the light of UFO.

Integration of upper and core layers: the upper and core layers are integrated mainly using inheritance links. This integration could be accomplished and verified within OLED during the conceptualization process. Thus, generalization links are added to relate concepts from upper and core layers (e.g., Legal Agent is a specialization of Agent).

Building the conceptual model of the domain layer: for building the conceptual model of the domain layer, the conceptual ontology patterns of the upper and core layers are applied either by extension or by analogy depending on their types. The recognition patterns are applied by extension for developing the static content of the domain layer. Meanwhile, the patterns as templates are applied by analogy with the legal rules for modeling the dynamic content of the domain layer. Furthermore, the domain layer’s content is checked in OLED to verify if it is built correctly.

3.5. Implementation

To validate the conceptual model and fulfill the non-functional requirements NFR3 and NFR4, CargO-S is implemented in OWL5

⁵
https://www.w3.org/TR/owl-features/

and SWRL6

⁶

https://www.w3.org/Submission/SWRL/

rules. The implementation phase consists of transforming the reference model of the ontology into an operational ontology represented in computationally tractable subsets of first-order logic, such as OWL or F-Logic (Brank et al., 2005). Besides, SWRL rules are used to express logic rules over OWL specifications (Brank et al., 2005).

3.6. Evaluation

The last phase of the ontology building process is ontology evaluation, which aims to: (1) evaluate the performance of CargO-S by employing the operational ontology in a practical application and (2) assess its semantic accuracy by computing the structural properties of the ontology.

4. Building the reference model of CargO-S

For building the reference model of CargO-S, the pattern-oriented top-down approach (see Section 3.4) is applied. In the following, we present the conceptual ontology patterns selected and reused from UFO and UFO-L for building the upper, core, and domain layers. The concepts, relations and axioms of the conceptual ontology patterns are represented in OntoUML using the modeling primitives of UFO.

4.1. Upper layer

For building the upper layer, the conceptual ontology patterns are selected and reused from UFO, mainly UFO-C and UFO-B. The upper layer is composed of three main DIOPs defined as pattern recognition: Substance, Event and Moment. In this layer, the ontology patterns implement the requirements at the most abstract granularity level.

Substance. The Substance ontology pattern, depicted in Fig. 8, implements the functional requirement FR1. This pattern is selected and reused from UFO-C (Guizzardi et al., 2008b). Two main CQs are addressed for this pattern: (CQ1) How are agents and objects structured? (CQ2) What categories are defined by normative description?

Substance can be Agent or Object. Agent may play a role such as carrier or shipper. Normative_Description (e.g., conventions of carriage of goods by sea), which defines one or more rules/norms, is a Social_Object recognized by at least one Social_Agent.

Fig. 8.

Substance ontology pattern (adapted from Guizzardi et al., 2008b).

Event. The Event ontology pattern, depicted in Fig. 9, implements the functional requirement FR2. This patterns is selected and reused from UFO-B (Guizzardi, 2013). Five CQs are addressed for this pattern: (CQ1) How are events structured? (CQ2) At what time points events begin and end (CQ3)? At which location events occur? (CQ4) Exist any events that cause the occurrence of other events? (CQ5) Exist any situation that triggers the occurrence of events?

Event, which can be Atomic_Event or Complex_Event, may be composed of other events (Guizzardi, 2013). While Atomic_Event has no proper parts, Complex_Event is aggregation of at least two disjoint Event. An Event, which is triggered by a Situation, may cause the occurrence of other events. Event occurs in a Spatial_Location, and begins and ends at Time_Point.

Fig. 9.

Event ontology pattern (adapted from Guizzardi, 2013).

Moment. The Moment ontology pattern, depicted in Fig. 10, implements the functional requirements FR3 and FR5. This pattern is selected and reused from UFO-C (Guizzardi et al., 2008b). Four main CQs are addressed for this pattern: (CQ1) How are relators organized? (CQ2) What event is the basis of relators? (CQ3) How are moments organized? (CQ4) What moments compose social relators?

Social_Relator is a Relator composed of two or more Externally_Dependent_Social_Moment: Social_Claim such as Right and Social_Commitment such as Duty.

Fig. 10.

Moment ontology pattern (adapted from Guizzardi et al., 2008b).

4.2. Core layer

For building the core layer, the conceptual ontology patterns are selected and reused from UFO-L (Griffo, 2018). In this layer, the ontology patterns implement the requirements at the domain-dependent level (legal domain). Thus, different DDOPs are defined as recognition and template patterns.

4.3. Recognition patterns

Legal_Substance. This pattern, depicted in Fig. 11, implements the functional requirement FR1. Three main CQs are addressed for selecting this pattern: (CQ1) How are legal agents structured? (CQ2) How are legal objects structured? (CQ3) What categories are defined by legal normative description? Legal_Substance is specified into Legal_Agent and Legal_Object.

Legal_Normative_Description is a Legal_Object having as content Legal_Norm. It defines Legal_Agent, Legal_Role, and Legal_Object.

Fig. 11.

Legal_Substance ontology pattern (adapted from Griffo, 2018).

Legally_Defined_Event. This pattern, depicted in Fig. 12, implements the functional requirement FR2. It represents a specialization of the Event pattern in the legal domain. Five main CQs are addressed for modeling this pattern: (CQ1) How legal events are structured? (CQ2) At what time points, legal events begin and end (CQ3) At which location legal events occur? (CQ4) Exist any legal events that cause the occurrence of other legal events? (CQ5) Exist any legal situation that triggers the occurrence of legal events?

Fig. 12.

Legally_Defined_Event ontology pattern.

Legal_Relator. This pattern, depicted in Fig. 13, implements the functional requirement FR3. Three main CQs are addressed for selecting this pattern: (CQ1) How are legal relators organized? (CQ2) What events are the basis of legal relators? (CQ3) What legal moments compose legal relators?

Legal_Relator, which is grounded by Legally_Defined_Event and defined by Legal_Rule, is a specialization of Social_Relator (Griffo, 2018). A Simple_Legal_Relator is specialized into different relators such as Right_Duty_Relator, Power_Subjection_Relator, and Disability_Immunity_Relator.

Fig. 13.

Legal_Relator ontology pattern (adapted from Griffo, 2018).

Externally_Dependent_Legal_Moment. This pattern, depicted in Fig. 14, implements the functional requirement FR5. Two main CQs are addressed in this pattern: (CQ1) How are legal moments organized? (CQ2) Are there correlations between legal moments?

In Externally_Dependent_Legal_Moment pattern, different legal moments are selected (Griffo et al., 2018): Right (i.e., legal moment in which one may demand from another the performance of a certain conduct), Right_to_an_Action, Duty (converse moment of Right), Duty_to_Act, Legal_Power (i.e., ability to act to a power holder), Legal_Subjection (converse moment of Legal_Power), Disability (i.e., no power to create, change or extinguish a legal moment), and Immunity (converse moment of Disability).

Fig. 14.

Externally_Dependent_Legal_Moment ontology pattern (adapted from Griffo, 2018).

4.4. Template patterns

In the core layer, three legal relators patterns are reused from UFO-L as template patterns (Griffo, 2018): Right_Duty_to_an_Action_Relator (Fig. 3), Power_Subjection_Relator (Fig. 4), and Disability_Immunity_Relator (Fig. 5). The reuse process include also the list of competency questions that are addressed for these patterns in UFO-L. The template patterns implement the functional requirements FR4 and FR6.

Right_Duty_to_an_Action_Relator. It represents the relationship where the Right_Holder has the right to a positive action by the Duty_Holder (Griffo, 2018). For example, the carrier of goods by sea owes a duty to a shipper of cargo that the vessel is seaworthy at the start of the voyage.

Power_Subjection_Relator. It represents the relationship where the Power_Holder has the competence (or the legal power) to create (change, extinguish) a legal position or a situation in the face of Subjection_Holder (Griffo, 2018). For example, the shipper has the power to indemnify the carrier for the liability of loss or damage of goods.

Disability_Immunity_Relator. It represents the relationship where Disability_Holder has no competence to act, in the face of Immunity_Holder, to create, modify, or extinguish legal positions or situations. It is the negation of Power_Subjection_Relator. For example, the carrier has immunity to liability against the shipper for loss or damage of goods resulting from strikes.

4.5. Domain layer

The domain layer is the result of applying the conceptual ontology patterns of the upper and core layers. It is composed of static and dynamic content. The static content represents the definition of the basic concepts of the domain of discourse and their structure. The dynamic content represents the procedural aspect of the legal relationships among the different agents. The recognition ontology patterns of the upper and core layers are applied by extension for building the static content. Meanwhile, the template patterns of the core layer are applied by analogy with the legal rules for building the dynamic content. In the domain layer, the ontological requirements are implemented at the domain-dependent field-dependent level, which is represented by the Hague-Visby Rules.

4.5.1. Building the static content

Extension of Physical_Object. For fulfilling the functional requirement FR1, the following CQs are addressed regarding physical objects: (CQ1) What are the main physical objects in the domain of carriage of goods by sea? (CQ2) How are these objects structured?

Based on Article I and Article IV, three main physical objects in the domain of carriage of goods by sea are defined: Goods, Vessel and Article_of_Transport. These concepts are specified by extending Physical_Object (Fig. 15). Goods are specified into Wares, Merchandises and Articles. Ship is the vessel defined in this domain. Two main types of Article_of_Transport are identified: Container and Pallet. They are used to consolidate goods and are carried on the deck of ships.

Article I. (c) ‘Goods’ includes goods, wares, merchandise, and articles of every kind whatsoever […]

Article I. (d) ‘Ship’ means any vessel used for the carriage of goods by sea.

Article IV. (5)(c) […] a container, pallet or similar article of transport is used to consolidate goods […]

Fig. 15.

The physical objects in the domain layer of CargO-S.

Extension of Legal_Normative_Description. For realizing the functional requirement FR1 regarding legal objects, there is a need to represent the modeling aspects of legal normative descriptions. For this purpose, the main CQs addressed are: (CQ1) What are the main regulations in the domain of carriage of goods by sea? (CQ2) How are these regulations structured?

In the conventions of carriage of goods by sea such as Hague-Visby Rules, different rules regulate the legal agents, legal objects, and legal relations. Thereby, these conventions is considered as a specification of Legal_Normative_Description (Fig. 16). Besides, the bill of lading (Article I) is considered a non-negotiable legal document defined by the conventions of the carriage of goods by sea. It regulates the contract of carriage of goods and the relations between the legal agents. Thereby, it is also defined as a specification of Legal_Normative_Description.

Article I. (b) […] bill of lading relates to the carriage of goods by sea […] such bill of lading regulates the relations between a carrier and a holder of the same.

Fig. 16.

Extension of Legal_Normative_Description in the domain layer of CargO-S.

Extension of Legal_Role. For fulfilling the functional requirement FR1, regarding legal agents or roles in the domain of carriage of goods by sea, the following CQs are addressed: (CQ1) What are the main legal roles in the domain of carriage of goods by sea? (CQ2) How are these legal roles structured?

The Legal_Role is extended to specify the main legal roles in the domain of carriage of goods by sea (Fig. 17). Based on Article I, two main legal roles are identified: Carrier and Shipper. The legal role Consignee is out of the scope of this work (see Section 3.1).

Article I. (a) ‘Carrier’ who enters into a contract of carriage with a shipper […]

Fig. 17.

The legal roles in the domain layer of CargO-S.

Fig. 18.

The Carriage_of_Goods_by_Sea event in the domain layer of CargO-S.

Extension of Complex_Legally_Defined_Event and Atomic_Legally_Defined_Event. For fulfilling the requirement FR2, complex and atomic carriage of goods events need to be clearly defined in the domain layer. For this purpose, the following CQs are addressed: (CQ1) What are the main complex events occurring in the domain of carriage of goods by sea? (CQ2) How are the complex events structured? (CQ3) What are the main atomic events occurring in the domain of carriage of goods by sea? (CQ4) How are the atomic events structured? Based on Article I and Article II, Complex_Legally_Defined_Event is extended into Carriage_of_Goods_by_Sea (Fig. 18) which is composed of the following atomic events: Loading_Goods, Handling_Goods, Carrying_Goods, and Discharging_Goods. In this work, we do not consider the procedural aspect of these events and thereby the activities of loading, handling, carrying and discharging goods are not discussed.

Article I. (e) ‘Carriage of Goods’ covers the period from the time when the goods are loaded on to the time they are discharged from the ship.

Article II. […] under every contract of carriage of goods by sea the carrier in relation to the loading, handling, carriage and discharge of such goods […]

Extension of Legally_Defined_Event. For fulfilling the requirement FR2, causal and resultant events need to be specified. For this purpose, the following CQs are addressed: (CQ1) What are the causal legally defined events in the domain of carriage of goods by sea? (CQ2) How are the causal events structured (CQ3) What are the main resultant events in the domain of carriage of goods by sea? (CQ4) How are the resultant events structured? Based on Article IV, two main types of legal events are defined in the domain layer (Fig. 19): Loss_or_Damage_of_Goods as resultant events and Causal_Legally_Defined_Event as causal events. Examples of causal events are Deviation, Fire, Strike, and War.

Article IV. (2) […] loss or damage arising or resulting from: […] fire, strikes, act of War […]

Fig. 19.

The Loss_or_Damage_of_Goods event in the domain layer of CargO-S.

Extension of Legally_Defined_Situation. For accomplishing the functional requirement FR2, it is required to specify the legally defined situations. For this purpose, the following CQs are addressed: (CQ1) what are the main situations occurring in the domain of carriage of goods by sea? (CQ2) How are these situations structured?

Fig. 20.

Extension of Legally_Defined_Situation in the domain layer of CargO-S.

Besides causal events, situations may trigger the occurrence of loss or damage of goods (Article III and Article IV). In the domain layer, three main legally defined situations have been identified (Fig. 20): (1) Unseaworthiness is a situation that may occur before or at the beginning of the carriage (e.g., the ship is not seaworthy); (2) Inaccuracies (e.g., incorrectness in the marks, number, quantity, and weight of goods); (3) Negligence_Fault_Failure (e.g., negligence in ventilation of goods). Negligence_Fault_Failure is a situation representing a failure in the legal agents’ duties or obligations (Article 3, Section 8). In this work, we do not consider Negligence_Fault_Failure as an event that may brings about a situation of Unseaworthiness or Inaccuracies. We are concerned in Negligence_Fault_Failure as a situation triggering the loss or damage of goods during their carriage (after the departure of the ship) (Arness, 1972). Thus, we differentiate between two main types of Negligence_Fault_Failure: Negligence_Fault_Failure_in_Navigation_or_Management_of_Ship (e.g., collisions or violent contact with other perils of the sea, failure to secure the deep tank of the ship (Arness, 1972)) and Negligence_Fault_Failure_in_Care_and_Custody_of_Goods (e.g., negligence in heating or cooling of cargo, improper handling or stowage of goods (Arness, 1972)).

Article III. (5) […] loss, damages or expenses arising or resulting from inaccuracies […]

Article IV. (1). […] loss or damage arising or resulting from unseaworthiness […]

Article III. (8) […] loss or damage […] arising from negligence, fault, or failure in the duties and obligations […]

Extension of legal relators. For fulfilling the functional requirement FR3, it is necessary to specify the legal relationships and their hierarchical and mereological structure in the domain layer. For this purpose, the following CQs are proposed: (CQ1) what are the main relations in the domain of carriage of goods by sea? (CQ2) How are these relations structured?

Fig. 21.

Extension of Right_Duty_to_an_Action_Relator in the domain layer of CargO-S.

The following legal relators are specified in the domain layer (Fig. 21).

Contract_of_Carriage_of_Goods_by_Sea: the commitment to carriage of goods is a contractual obligation between the carrier and the shipper (Article I). Since the carriage covers the period from when the goods are loaded to the time they are discharged from the ship, this commitment includes the following obligations: accuracy, care, custody, and carriage of cargo and seaworthiness of the ship (Article III and Article IV).

Article I. (b) ‘Contract of carriage’ applied only to contracts of carriage covered by bill of lading […] regulates the relations between a carrier and a holder of the same.

Right_Duty_to_Accuracy: accuracy is defined as a legal obligation to the shipper against the carrier. The shipper has to guarantee to the carrier, at the time of shipment, the accuracy of the marks, number, quantity, and weight of the goods (Article III (5)).

Article III. (5) The shipper shall be deemed to have guaranteed to the carrier the accuracy at the time of shipment […]

Right_Duty_to_Seaworthiness: seaworthiness is declared as a legal obligation and a primary carrier’s commitment within the conventions of the international maritime law. The carrier is responsible before and at the beginning of the voyage to make the ship seaworthy and secure that the ship is properly manned, equipped, and supplied (Article III (1)).

Article III. (1) The carrier shall be bound before and at the beginning of the voyage to […] make the ship seaworthy […]

Right_Duty_to_Care_and_Custody_of_Cargo: care and custody of goods is declared as a major legal obligation of the carrier. The carrier is responsible of the care and custody of goods during the carriage period (Article III (2)).

Article III. (2) […] the carrier shall properly and carefully load, handle, stow, carry, keep, care for, and discharge the goods carried.

Right_Duty_to_Carriage_of_Goods: the carriage of goods is a basic obligation relations between the carrier and the shipper. The carrier, as duty holder, is responsible of loading, handling, carrying and discharging goods (Article III (2)).

Power_Subjection_to_Liabilities: power to liability against a liable agent is defined in different rules in Hague-Visby. Generally, the shipper has a power to liability against the carrier under the contract of carriage of goods. An exception is the power of the carrier against the shipper in case of inaccuracies (Article III (5)).

Article III. (5) […] the shipper shall indemnify the carrier against all loss, damages and expenses arising or resulting from inaccuracies […]

Right_Duty_to_Indemnity: indemnity is defined as a recompense for loss or damage of goods. It is a commitment to the liable agent against the claimant of the damage (Article III (5)).

Disability_Immunity_to_Liabilities: immunity to liabilities is declared in different rules of Hague-Visby. It can be applied in different situations such as, Article IV (1), where the carrier is exempted from liability of loss or damage of goods because of unseaworthiness unless he was a participant.

Article IV. (1) Neither the carrier nor the ship shall be liable for loss or damage arising or resulting from unseaworthiness unless caused by want of due diligence on the part of the carrier to make the ship seaworthy.

4.5.2. Building the dynamic content

The dynamic content of the domain layer aims to represent the procedural aspect of the Hague-Visby Rules. These rules argue mainly the rights, duties, liabilities, and immunities of legal agents. Thus, the following legal relators are described: Right_Duty_to_Carriage_of_Goods, Right_Duty_to_Indemnity, Power_Subjection_to_Liabilities, and Disability_Immunity_to_Liabilities. Aiming to fulfill the requirements of the targeted ontology, specifically FR4 and FR6, specific competency questions are defined for each legal relator by considering and specializing those determined in UFO-L (see Section 2). These questions address mainly the representation of the procedural aspect of the legal relators. Besides, general competency questions addressing the purpose of each legal relator are also addressed.

Right_Duty_to_Carriage_of_Goods. This legal relator (Fig. 22) is defined based on the application by analogy of Right_duty_to_an_Action_Relator template pattern (Fig. 3) with Article I and Article II. The main CQs addressed for modeling this legal relator are: (CQ1) Who is the agent responsible for the carriage of goods? (CQ2) To whom the profit of the carriage refers? The specialized CQs are: (CQ3) Which legal roles are involved in the right-duty to carriage? (CQ4) What are the legal moments that make up the right-duty to carriage? (CQ5) Who are the holders of each moment? (CQ6) On whom is each legal moment externally dependent? (CQ7) What is the event that grounds the right-duty to carriage? (CQ8) Is there a legal rule that defines the right-duty to carriage?

Fig. 22.

The procedural modeling of Right_Duty_to_Carriage_of_Goods.

Right_Duty_to_Carriage_of_Goods mediates a Carrier as Duty_Holder, and a Shipper as Right_Holder. It is grounded on the Carriage_of_Goods_by_Sea event and defined by a Bill_of_Lading. The legal relator is composed of the following correlated legal moments: Right_to_Load and Duty_to_Load, Right_to_Carry and Duty_to_Carry, Right_to_Handle and Duty_to_Handle, and Right_to_Discharge and Duty_to_Discharge (Article II).

Article II. […] under every contract of carriage of goods by sea the carrier in relation to the loading, handling, carriage and discharge of such goods shall be subject to responsibilities and entitled to the rights […]

In this model, a material relation, named has a right to carriage against, is derived from the Right_Duty_to_Carriage_of_Goods legal relator and the corresponding mediation relations.

Power_Subjection_to_Liabilities. The Power_Subjection_to_Liabilities legal relator (Fig. 23) is specified in the context of Loss_or_Damage_of_Goods. The main CQs addressed for modeling this legal relator are: (CQ1) which legal rule is being violated, if any? (CQ2) What are the damage triggering situations, if any? (CQ3) Who are the involved agents in such cases? (CQ4) What are the damage-causing events, if any? (CQ5) Who are the active agents in such events? (CQ6) Who does the agent have the legal power? The specialized CQs are: (CQ7) Which legal roles are involved in the power-subjection to liabilities? (CQ8) What are the legal moments that make up the power-subjection to liabilities? (CQ9) Who are the holders of each moment? (CQ10) On whom is each legal moment externally dependent? (CQ11) What is the event that grounds the power-subjection to liabilities? (CQ12) Is there a legal rule that defines the power-subjection to liabilities? (CQ13) What is the legal relationship created, modified, or extinguished? (CQ14) Who are the owners of the created, modified, or extinct legal relationship?

Fig. 23.

The procedural modeling of Power_Subjection_to_Liabilities.

The causation of Loss_or_Damage_of_Goods defines the liability of the involved legal agents. Generally, the shipper indemnifies the carrier against the loss or damage of goods resulting from inaccuracies (Article III (5)). Thereby, the Carrier as Subjection_Holder has a power against the Shipper as Power_Holder.

Article III. (5) […] the shipper shall indemnify the carrier against all loss, damages and expenses arising or resulting from inaccuracies […]

Power_Subjection_to_Liabilities legal relator is grounded on Loss_or_Damage_of_Goods which is triggered by Inaccuracies having a part the Shipper. This relator creates, modifies or extinguishes minimum one Right_duty_to_Indemnity relator. The material relation has a legal power against is derived from Power_Subjection_to_Liabilities and the corresponding mediation relations.

Fig. 24.

The procedural modeling of Right_Duty_to_Indemnity.

Right_Duty_to_Indemnity. The main CQs addressed for modeling this relator (Fig. 24) are: (CQ1) which legal rule is being violated, if any? (CQ2) What are the damage triggering situations, if any? (CQ3) Who are the involved agents in such cases? (CQ4) What are the damage-causing events, if any? (CQ5) Who are the active agents in such events? (CQ6) Who is the liable agent? The specialized CQs are: (CQ7) Which legal roles are involved in the right-duty to indemnity relationship? (CQ8) What are the legal moments that make up this relationship? (CQ9) Who are the holders of each moment? (CQ10) On whom is each legal moment externally dependent? (CQ11) What is the event that grounds the right-duty to indemnity relationship? (CQ12) Is there a legal rule that defines this relationship?

Right_Duty_to_Indemnity is grounded on Loss_or_Damage_of_Goods. The latter is triggered by Inaccuracies having as part the Shipper. Thus, the Shipper practices and the Carrier suffers the Loss_or_Damage_of_Goods. In this model, the material relation has a right to indemnity against is derived from the Right_Duty_to_Indemnity relator and the corresponding mediation relations.

Disability_Immunity_to_Liabilities. This legal relator represents the denial of the Power_Subjection_to_Liabilities relator. It describes the situation where the Carrier or the Shipper have immunities to liabilities, which eliminate any competence to indemnify them in case of loss or damage of goods (Article IV). In the Hague-Visby Rules, different cases are specified based on the causation of the Loss_or_Damage_of_Goods event, such as the following:

Loss or damage of goods resulting from unseaworthiness (Article IV (1)): the carrier is not liable if he was not a participant.

Loss or damage of goods resulting from unanticipated causal events Article IV (2): the carrier is exempted from any responsibility.

Article IV. (2) Neither the carrier nor the ship shall be responsible for loss or damage arising or resulting from: […] fire, strikes, act of God, act of War […]

Loss or damage of goods arising from any cause without the act, fault, or neglect of the shipper Article IV (3): the shipper is exempted from liability if the damage is not a result of fault or negligence of obligations.

Article IV. (3). The shipper shall not be responsible for loss or damage sustained by the carrier or the ship arising or resulting from any cause without the act, fault or neglect of the shipper […]

In Fig. 25, the conceptual model of Disability_Immunity_to_Liabilities relator is illustrated. The main CQs addressed for this model are: (CQ1) which legal rule is being violated, if any? (CQ2) What are the damage triggering situations, if any? (CQ3) Who are the involved agents in such situations? (CQ4) What are the damage-causing events, if any? (CQ5) Who are the involved agents in such events? (CQ6) Who does the agent have the immunity to liability? The specialized CQs are: (CQ7) Which legal roles are involved in the disability-immunity to liabilities relationship? (CQ8) What are the legal moments that make up the disability-immunity to liabilities relationship? (CQ9) Who are the holders of each moment? (CQ10) On whom is each legal moment externally dependent? (CQ11) What is the event that grounds the disability-immunity to liabilities relationship? (CQ12) Is there a legal rule that defines the disability-immunity to liabilities relationship?

Disability_Immunity_to_Liabilities, which is grounded on Loss_or_Damage_of_Goods, is triggered by Unseaworthiness having no part the Carrier. In this model, the material relation has disability to liability against is derived from the Disability_Immunity_to_Liabilities relator and the corresponding mediation relations.

Fig. 25.

The procedural modeling of Disability_Immunity_to_Liabilities.

5. Ontology implementation

The implementation phase aims to validate the reference ontology by producing an operational version of CargO-S. The ontology environment OLED (Guerson et al., 2015) provides the transformation from reference to operational ontology by generating the OWL code. The code generator maps OntoUML classes, associations, and attributes to OWL classes, object properties, and data properties. It considers generalization sets and its disjointness properties plus model cardinalities. Although the code generator considers the domain constraints, cardinalities, transitivity of material and parthood relations as SWRL rules. Besides, we present a list of formal rules describing the procedural aspect of the legal relators representing the legal rules. To formalize the rules, we have used SWRL as a formal rule language. The SWRL rules will be employed as a rule base for decision support purposes. For the metrics, CargO-S comprises 122 classes, 209 SubClassOf relations, and 412 object properties. In this section, we present an excerpt of the operational version of CargO-S, which is accessible and manageable by ontology editors such as Protégé.7

⁷
https://protege.stanford.edu/

In the following, example of classes, properties, and SWRL rules are represented.

Loss_or_Damage_of_Goods. This legal event (Fig. 26) grounds three main legal relators: Power_Subjection_to_Liabilities (Fig. 23), Right_Duty_to_Indemnity (Fig. 24), and Disability_Immunity_to_Liabilities (Fig. 25).

Fig. 26.

Loss_or_Damage_of_Goods represented in Protégé.

Right_Duty_to_Indemnity. The Right_Duty_to_Indemnity legal relator (Fig. 27) is composed of two legal moments Duty_to_Indemnity and Right_to_Indemnity, and grounded on Loss_or_Damage_of_Goods. This legal relator is defined by a Legal_Rule and mediates a Carrier and a Shipper.

Fig. 27.

Right_Duty_to_Indemnity represented in Protégé.

Object properties. In CargO-S, object properties are defined based on the relations specified in OntoUML. Different types of relations have been used such as, Formal, Mediation, Characterization, Material, Derivation, Part-Whole, and ComponentOf. The characteristics and constraints of the relations are defined in OntoUML.8

⁸

https://ontouml.readthedocs.io/en/latest/, last visited January 17, 2021.

For example, ComponentOf relations are parthood relations between two complexes. It is a directed relation, having the following binary properties: Reflexivity: No, Transitivity: No, Symmetry: No, and Cyclicity: No. In Fig. 28 , this relation is depicted as implemented in Protégé.

Fig. 28.

ComponentOf represented in Protégé.

SWRL rules. For decision support purposes, there is a need to formalize, based on CargO-S, the procedural aspect of the legal rules of the Hague-Visby convention. To achieve the goal, two main requirements are identified (Gordon et al., 2009): (1) a rule language based on a precise and rigorous semantics, which allows for correctly computing the legal effects that should follow from a set of legal rules; (2) the application of the isomorphism principle, stated by Bench-Capon (Bench-Capon and Coenen, 1992), to create a well-defined correspondence between the rules in the formal model and the units of natural language text that express the rules in the original legal sources as sections of the legislation. Regarding the rule language, we choose SWRL, a rule interchange format that combines ontologies represented in OWL with RuleML. SWRL extends OWL axioms to include (monotonic) Horn-like rules. It is the only approach that gathers ontology and rules in product development (Fiorentini et al., 2010). Users are permitted to write rules that can be expressed in terms of OWL concepts and reason about OWL individuals (O’Connor et al., 2008).

In CargO-S, SWRL rules are generated either automatically or manually. They are represented by obligation rules in the following form: IF condition (operative facts) THEN conclusion (legal effect). Concerning the automatically generated rules, they are obtained based on the material relations defined in the procedural conceptual models of the legal relators. As example, we consider the following SWRL rule generated from the conceptual model of Right_Duty_to_Indemnity (Fig. 24).

Concerning the formal rules that are implemented manually, we have followed the isomorphism principle for connecting the textual rules with formalized rules. The formalization process is performed based on the dynamic modeling of the legal rules in CargO-S. As example, we consider the following SWRL rule generated from the conceptual model of Right_Duty_to_Indemnity (Fig. 24).

As a result, we obtained for Right_Duty_to_Indemnity legal relator two chained formal SWRL rules where the former’s condition is the latter’s consequence. The same process is applied for all the legal relators representing the procedural aspects of the legal rules in the domain layer of CargO-S.

6. Ontology evaluation

In the literature, the most commonly known approaches for the evaluation of ontologies are (Brank et al., 2005; Fernandez et al., 2009): (1) gold standard-based which aims to compare the developed ontology with a previously created reference ontology, (2) task-based which evaluates ontologies, that are intended for particular applications, according to their performance, and (3) structure-based which computes structure-based properties such as the size and the complexity of a given ontology. Since CargO-S is an application-based ontology, it is evaluated using a dual evaluation approach: task-based that evaluates the usability of CargO-S in a practical application and structure-based that assesses the semantic accuracy of CargO-S. The latter approach is recommended as an efficient approach for evaluating the developed ontologies (Dellschaft and Staab, 2008). We eliminated the gold standard-based approach due to the difficulty of finding a reference ontology that should be created under similar conditions with similar goals of CargO-S.

6.1. Task-based evaluation: Practical application

This section evaluates CargO-S being an application-based ontology by embedding it in a practical application. Seeking the traceability of goods, two main tasks are identified: (1) data retrieval using basic SPARQL queries and (2) decision support using SWRL rules. To achieve the goal, CargO-S is enriched with a set of instances provided by the domain experts.

6.1.1. Data retrieval using SPARQL queries

A set of basic SPARQL queries are specified and executed aiming to answer the set of competency questions. These queries tend to retrieve data related to the carriage of goods, such as the origin and destination of goods, the contracts of carriage of goods and the involved legal agents, and the events and situation occurred during transportation. In the following, examples of basic SPARQL queries are introduced. $Q 1$ (Fig. 29) retrieves the events of carriage of goods, the countries of origin and destination, the ports, and the dates of shipment and discharge. In Fig. 30, excerpt of $Q 1$ results is illustrated. $Q 2$ (Fig. 31) returns the contracts of carriages of goods, and the involved legal agents. In Fig. 32, excerpt of $Q 2$ results is illustrated. $Q 3$ (Fig. 33) identifies the events of loss or damage of goods and the situations triggering these events. In Fig. 34, excerpt of $Q 3$ results is illustrated.

Fig. 29.

The SPARQL query $Q 1$ .

Fig. 30.

Excerpt of $Q 1$ results.

Fig. 31.

The SPARQL query $Q 2$ .

Fig. 32.

Excerpt of $Q 2$ results.

Fig. 33.

The SPARQL query $Q 3$ .

Fig. 34.

Excerpt of $Q 3$ results.

6.1.2. Decision support using SWRL rules

During the transportation of goods, loss or damage of goods may result from causal events or triggering situations. In this context, there is a need to define the liability and responsibility of the involved legal agents. This task is performed, based on the SWRL rules, as a lightweight decision support task. The reasoner $HermiT$ is used for the inference process depicted in Fig. 35. The inference engine matches the SWRL rules (specifically the antecedent or the condition of the rule) stored in a rule base with the given facts. The matched rules are executed to generate legal effects. In this section, we discuss two different cases: the damage of cargo arising out of triggering situations (e.g., negligence in the care and custody of goods), and the loss of goods resulted from legally defined causal events (e.g., strike).

Fig. 35.

The inference process.

Triggering situation: Negligence in the care and custody of cargo. Given an event of carriage of lychees operated by the carrier Service Capricorne and the shipper Lacour Exotics. For respecting the obligations of care and custody of goods, the lychees are consolidated in containers under controlled temperature. During transportation, the required temperature has not been respected causing the damage of fruits. Based on the given facts (Fig. 36), the following chained SWRL rules are inferred. Thereby, different legal effects are computed and generated following the execution of these rules.

As a result, we obtained that the shipper Lacour Exotics has a right to indemnity against Service Capricorne. In Fig. 37 and Fig. 38, excerpts of the generated legal effects are depicted.

Fig. 36.

Excerpt of the given facts represented in Protégé.

Fig. 37.

Excerpt of the generated legal effects (Shipper) represented in Protégé.

Fig. 38.

Excerpt of the generated legal effects (Right_Duty_to_Indemnity) represented in Protégé.

Legally defined causal events: Strikes. Given an event of carriage of books operated by the carrier Pen Duick and the shipper INTERNATIONAL LOG BOOK. At the destination port Le Havre, and before discharging the goods, a strike has occurred on the ship. As a result, the containers have been looted at the terminal. Based on the given facts (Fig. 39), the following chained SWRL rules are inferred. Thereby, different legal effects are computed and generated following the execution of these rules.

As a result, we obtained that the carrier Pen Duick has an immunity to liability and the shipper INTERNATIONAL LOG BOOK has no power to liability against the carrier. In Fig. 40 and Fig. 41, excerpts of the generated legal effects are depicted.

Fig. 39.

Excerpt of the given facts represented in Protégé.

Fig. 40.

Excerpt of the generated legal effects (Shipper) represented in Protégé.

Fig. 41.

Excerpt of the generated legal effects (Disability_Immunity_to_Liabilities) represented in Protégé.

6.2. Structure-based evaluation: Semantic accuracy

The structure-based evaluation is an automatic ontology evaluation which aims to assess the structural quality of CargO-S. Several measures have been recognized in the literature, such as Knowledge coverage and popularity measures (e.g., number of classes and number of properties) and structural measures (e.g., maximum depth, average depth, depth variance, etc.) (Fernandez et al., 2009). The application of these measures relies on the assumption that is a richly populated ontology, with high depth and breadth variance is more likely to provide reliable semantic content. The Knowledge coverage and popularity measures, which are commonly used in the ontology evaluation literature, do not show a significant relationship with the ontological accuracy (Sanchez et al., 2015). However, the structural measures are positively correlated with the semantic accuracy of the knowledge modeled in the ontology (Sanchez et al., 2015).

In CargO-S, we quantified some structural measures, which can be used to predict the ontologies with the best reliability (Sanchez et al., 2015), by considering the taxonomic structure of the ontology graph:

Maximum depth: represents the length of the longest taxonomic branch in the ontology. It is measured as the number of concepts from the root node to the leaves of the taxonomy. In CargO-S, $maximum depth = 8$ .

An example of a taxonomic long path is:

${Right_Duty_to_Indemnity \to Right_Duty_to_an_Action_Relator \to Right_Duty_Relator \to Simple_Legal_Relator \to Legal_Relator \to Social_Relator \to Relator \to Moment \to owl : Thing}$ .

Average depth: is the average length of all the taxonomic branches. In CargO-S, $average depth = 4.14$ .

Depth variance: is the dispersion with respect to the average depth, computed as the application of the standard mathematical variance to the depths of ontology nodes (see Equation (1)). In CargO-S, $depth variance = 5.37$ . $\begin{matrix} (1) & Depth_Variance (CargO - S) = \frac{\sum_{c_{i} \in C} {(depth (c_{i}) - \overline{depth (C)})}^{2}}{| C |}, \end{matrix}$ where C is the set of concepts in CargO-S, $| C |$ is the cardinality of C excluding the Root ( $owl : Thing$ ), $depth (c_{i})$ is the shortest path between each concept $c_{i} \in C$ and the taxonomic Root node, and $\overline{depth (C)}$ is the average of depths $\forall c_{i} \in C$ .

Maximum breadth: represents the width of the taxonomic level of the ontology with the largest number of concepts. In CargO-S, $maximum breadth = 29$ (for $level = 7$ ).

Average breadth: is the average breadth of all the ontology levels. In CargO-S, $average breadth = 13.55$ .

Breadth variance: is the dispersion with respect to the average breadth, computed as the application of the standard mathematical variance to the breadths of ontology levels (see Equation (2)). In CargO-S, $breadth variance = 22.76$ . $\begin{matrix} (2) & Breadth_Variance (CargO - S) = \frac{\sum_{l_{i} \in L} {(breadth (l_{i}) - \overline{breadth (L)})}^{2}}{| L |}, \end{matrix}$ where L is the set of levels in CargO-S, $| L |$ is the cardinality of levels, $breadth (l_{i})$ is the width of each taxonomic level, and $\overline{breadth (L)}$ is the average of breadths $\forall l_{i} \in L$ .

Based on the assessed structural properties, we conclude that the depth and breadth variance, which are limited by the maximum depth and breadth of the taxonomy, are considered high values regarding the average depth and breadth values. Thereby, CargO-S is considered a richly populated ontology where the concepts are dispersed in an unbalanced taxonomy.

7. Related work: LKIF-core

In this section, we outline LKIF-Core, a legal core ontology proposed by Hoekstra and his colleagues (Hoekstra et al., 2009). LKIF-Core is an operational ontology implemented in OWL.9

⁹
https://github.com/RinkeHoekstra/lkif-core, last visited January 14, 2021.

It is composed of 15 modules categorized into three main categories, each of which describes a set of closely related concepts from both legal and commonsense domains (Hoekstra et al., 2009). First, the Abstract Concepts are defined in five closely related modules: top, place, mereology, time, and spacetime. Most classes of this category are borrowed from LRI-Core10

¹⁰

http://www.leibnizcenter.org/previous-projects/lricore, last visited January 18, 2021.

). Second, the Basic Concepts are distributed across four modules: process, role, action, and expression. Finally, the Legal Concepts, which represent the basic clusters, are extended by three modules establishing the legal ontology: legal action, legal role, and norm. LKIF-Core considers the representation of normative knowledge which is derived from (Kelsen, 1991). Kelsen considers norms as descriptions of an ideal world which can be seen as the basis of deontic logic (Breuker et al., 2004). Deontic logic is the basis for the normative reasoning in the AI and Law research domain by providing interpretations for the terms obligation, prohibition and permission (Breuker et al., 2004).

LKIF-Core has been reused for building different legal domain ontologies such as OPJK (Ontology of Professional Judicial Knowledge) (Casellas, 2008) and a domain ontology to fill the gap between legal texts and rules (Palmirani et al., 2012). OPJK is a legal ontology developed to map junior judges’ questions to a set of stored frequently asked questions. The purpose of OPJK is to search and index for a web-based application. The ontology contains domain-specific knowledge and professional knowledge, gathered by experience from the practice during on-call periods with semi-structured interviews. The main top classes of OPJK are Role, Agent, Document, Process, and Act. In (Palmirani et al., 2012), to fill the gap between legal texts and rules, an ontology for modeling and defining macro-concepts specific for the legal domain is developed and expressed in LKIF-Core. The approach is applied to a fragment of the legal norms of the US copyright domain. In such ontologies, the structural representation of the legal domain is maintained. However, there is a lack in a representation of the procedural aspect of the legal rules. This is because the existent approaches have failed to support legal relations and capture the legal roles played in the context of these relations (Griffo et al., 2018).

8. Discussion

This study’s main contribution is the development of a well-founded legal domain ontology named CargO-S, representing the domain of carriage of goods by sea. CargO-S is targeted to be used as a knowledge base for the traceability of goods in logistic corridors. We applied a pattern-oriented approach supported by ontology reuse, layering, and grounding processes. To simplify the building process, the structure of CargO-S is divided into three layers: upper, core, and domain. For building upper and core, conceptual ontology patterns are selected and reused from the unified foundational ontology UFO and the legal core ontology UFO-L. Furthermore, these patterns are applied, either by extension or analogy, depending on their types, for building the domain layer.

The second contribution is that we differentiated between the reference model and the operational version of CargO-S. The proposed approach aimed first to develop the conceptual model of CargO-S by reusing and applying conceptual ontology patterns from UFO and UFO-L. Furthermore, the conceptual model is implemented in OWL and SWRL as an operational ontology. Besides, we separated the structural knowledge from the domain knowledge. In CargO-S, the representation of the hierarchy of concepts is distinguished from modeling the legal rules’ procedural aspects. This distinction will support the extensibility of the ontology to include future aspects.

Finally, we assume that by reusing UFO-L’s legal relators’ patterns and applying them by analogy with the legal rules, we obtained a richly populated ontology representing the procedural aspects of these rules. This result is considerable for employing our ontology in decision support purposes and is difficult to achieve by reusing other legal core ontologies such as LKIF-Core. Our assumption is based on the deficiency of representing “legal procedures” in LKIF-Core where concepts such as Change, Process, Action, Creation, and Reaction are used for this purpose. The notion of “legal relators” describing “legal procedures” is only specified in UFO-L. This lack has been recognized in the ontology engineering community, specifically in (Visser and Bench-Capon, 1997) where the authors affirm that most ontologies did not have an adequate solution for legal procedures mainly because of the difficulty to find a language to express knowledge in a declarative way.

Despite the contributions, this study has two principal limitations in (1) conceptualization and (2) decision support. Concerning the conceptualization, CargO-S does not cover the full range of core legal concepts. It focuses on modeling legal positions and relations and requires a representation of the normative knowledge considered in LKIF-Core. This deficiency is due to the purpose of CargO-S, which is mainly the definition of liabilities and responsibilities. Besides, essential legal roles such as consignee, master, and agent of the carrier are not represented due to the application domain’s limitation. In Hague-Visby Rules, the mentioned legal roles are not defined explicitly. Meanwhile, they are inherent in the definition of liabilities in the loss or damage of goods (Hamburg Rules). The second limitation is related to the use of SWRL for decision support. SWRL suffers from a lack of non-monotonic features (since it works under the Open-World Assumption (OWA)) (Fiorentini et al., 2010); Thus, it does not support the negation as failure. Besides, using SWRL makes difficult the representation of complex rules (i.e., contain exceptions or contradictions) (Fortineau et al., 2012) using conjunctions (∧). Disjunctions (∨) and defeasible logic, which are unsupported in SWRL, are required.

9. Conclusion

Building legal domain ontologies is a challenging task in the ontology engineering domain. Ontology builders face obstacles such as the complexity of the legal knowledge, the implicit definition of essential legal concepts, and the difficulty of applying existent ontology engineering methodologies. Besides, the legal ontologies are specific since they have to cover a wide range of common-sense concepts.

This work demonstrated that reusing ontology patterns from existent validated ontologies, sustained by ontology-driven conceptual modeling, is a practical approach that helps overcome the difficulties of building well-founded legal domain ontologies. This study applied a pattern-oriented approach supported by ontology reuse, grounding, and layering processes. The approach is based mainly on reusing conceptual ontology patterns from the foundational ontology UFO and the legal core ontology UFO-L. The main benefit of ontology layering and reuse processes is that they simplified the development process and enriched the ontology with axiomatizations inherited from UFO and UFO-L. The ontologically-driven conceptual modeling language OntoUML is applied for representing the categories of the targeted ontology in the light of UFO. As a result, we obtained CargO-S, a well-founded legal domain ontology with a significant ontological expressiveness. The validation and evaluation of CargO-S proved the ontology’s effectiveness in data querying and (lightweight) decision support purposes. Besides, the assessment of the semantic properties of CargO-S verified its semantic accuracy and richness.

In future work, we will extend the scope of the current version of CargO-S to cover other conventions of carriage of goods by sea, such as the Hamburg Rules (Hamburg, 1978) aiming to enrich the domain layer. Thereby, we will consider additional legal roles, such as the consignee and relationships, such as the delivery of goods and the associated responsibilities. The enrichment process requires a definition of an ontology re-engineering phase. In this study, we gave a lightweight, practical application for the validation of CargO-S. We will enhance the framework to include more detailed scenarios provided by the domain experts in further works. For this purpose, a specification task is needed to enable an interactive validation with the domain experts. Finally, to overcome the limitations of SWRL in decision support, Logic Programming (Kowalski, 1974) will be envisaged.

Footnotes

Acknowledgements

This work is supported by the European Regional Development Fund (ERDF; Grant No: HN0002134). The authors gratefully acknowledge Tatyana Poletaeva for the valuable discussions, especially in UFO. We are also thankful to the Institute of International Transport Law (IDIT) for the support.

References

Alexy, R. (2009). A Theory of Constitutional Rights. USA: Oxford University Press.

Arness, F. (1972). Error in navigation or management of vessels: A definitional dilemma. William and Mary Law Review, 13.

Ashley, K. (2009). Ontological requirements for anological teleological, and hypothetical legal reasoning. In Proceedings of the 12th International Conference on AI and Law (ICAIL).

Barcelos, P.P., Santos, V.A., Silva, F.B., Monteiro, M.E. & Garcia, A.S. (2013). An automated transformation from OntoUML to OWL and SWRL. In Proceedings of the 6th Seminar on Ontology Research, Brazil.

Bench-Capon, T. & Coenen, F. (1992). Isomorphism and legal knowledge based systems. Artificial Intelligence and Law, 1(1), 65–86. doi:10.1007/BF00118479.

Benevides, A.B., Guizzardi, G., BragaJoao, B.F. & Almeida, P.A. (2009). Assessing modal aspects of OntoUML conceptual models in alloy. In ER 2009: Advances in Conceptual Modeling – Challenging Perspectives (pp. 55–64).

Benjamins, V.R., Casanovas, P., Breuker, J. & Gangemi, A. (2005). Law and the Semantic Web: Legal Ontologies, Methodologies, Legal Information Retrieval and Applications. Springer.

Blomqvist, E., Gangemi, A. & Presutti, V. (2009). Experiments on pattern-based ontology design. In Proceedings of K-CAP (pp. 41–48). USA: ACM.

Borgo, S. & Masolo, C. (2010). Ontological foundations of DOLCE. In Theory and Applications of Ontology: Computer Applications. Springer.

10.

Brank, J., Grobelnik, M. & Mladenić, D. (2005). A survey of ontology evaluation techniques. In Proceedings of the Conference on Data Mining and Data Warehouses (SiKDD 2005).

11.

Breuker, J., Valente, A. & Wikels, R. (2004). Legal ontologies in knowledge engineering and information management. Artificial Intelligence and Law, 12, 241–277. doi:10.1007/s10506-006-0002-1.

12.

Casellas, N. (2008). Modelling legal knowledge through ontologies. OPJK: The ontology of professional judicial knowledge. PhD thesis, Universitat Autonoma de Barcelona.

13.

Daduna, J., Hunke, K. & Prause, G. (2012). Logistics corridors and short sea shipping in the Baltic Sea area. In Proceedings of the International Research Conference on Short Sea Shipping, Portugal.

14.

Dellschaft, K. & Staab, S. (2008). Strategies for the evaluation of ontology learning. In Proceedings of the 2008 Conference on Ontology Learning and Population: Bridging the Gap Between Text and Knowledge. Frontiers in Artificial Intelligence and Applications (Vol. 167, pp. 253–272).

15.

Dove, I. (1996). Legal expert systems: The end of jurisprudence? The Journal of Legal Studies in Business, 5.

16.

El Ghosh, M. & Abdulrab, H. (2019). The application of ODCM for building well-founded legal domain ontologies: A case study in the domain of carriage of goods by sea. In Proceedings of the Seventeenth International Conference on Artificial Intelligence and Law, ICAIL 2019 (pp. 204–208).

17.

Falbo, R. (2014). SABiO: Systematic Approach for Building Ontologies.

18.

Falbo, R., Baiao, F., Lopes, M. & Guizzardi, G. (2010). The role of foundational ontologies for domain ontology engineering: An industrial case study in the domain of oil and gas exploration and production. International Journal of Information System Modeling and Design, 1, 1–22. doi:10.4018/jismd.2010040101.

19.

Falbo, R., Barcellos, M.P., Nardi, J.C. & Guizzardi, G. (2013b). Organizing ontology design patterns as ontology pattern languages. In 10th Extended Semantic Web Conference, France.

20.

Falbo, R., Guizzardi, G., Gangemi, A. & Presutti, V. (2013a). Ontology patterns: Clarifying concepts and terminology. In Proceedings of WOP 2013 (Vol. 1188, pp. 14–26). Germany: CEUR-WS.

21.

Fernandez, M., Overbeeke, C., Sabou, M. & Motta, E. (2009). What makes a good ontology? A case-study in fine-grained knowledge reuse. In The Semantic Web (pp. 61–75). Berlin, Heidelberg: Springer. doi:10.1007/978-3-642-10871-6_5.

22.

Fiorentini, X., Rachuri, S., Suh, H., Lee, J. & Sriram, R. (2010). An analysis of description logic augmented with domain rules for the development of product models. Journal of Computing and Information Science in Engineering, 10, 1–13. doi:10.1115/1.3385794.

23.

Fortineau, V., Paviot, T., Sidney, L. & Lamouri, S. (2012). Swrl as a rule language for ontology-based models in power plant design. In Product Lifecycle Management. Towards Knowledge-Rich Enterprises. PLM (pp. 588–597). Berlin, Heidelberg: Springer. doi:10.1007/978-3-642-35758-9_53.

24.

Francesconi, E. & Tiscornia, D. (2008). Building semantic resources for legislative drafting: The DALOS project. In Computable Models of the Law. LNCS (Vol. 4884, pp. 56–70). Berlin, Heidelberg: Springer.

25.

Gangemi, A. (2005). Ontology design patterns for semantic web content. In International Semantic Web Conference. LNCS (Vol. 3729, pp. 262–276). USA: Springer, Heidelberg.

26.

Gangemi, A. (2007). Design patterns for legal ontology constructions. In LOAIT.

27.

Gangemi, A. & Presutti, V. (2009). Ontology Design Patterns. Heidelberg: Springer.

28.

Gardner, A. (1987). Book review. In An Artificial Intelligence Approach to Legal Reasoning. LNCS (Vol. 4884, pp. 223–233). MIT Press.

29.

Gordon, T., Governatori, G. & Rotolo, A. (2009). Rules and norms: Requirements for rule interchange languages in the legal domain. In RuleML 2009: Rule Interchange and Applications. LNCS (Vol. 5858, pp. 282–296). Berlin, Heidelberg: Springer.

30.

Griffo, C. (2018). UFO-L, a core ontology of legal aspects building under the perspective of legal relations. PhD thesis, Federal University of Espirito Santo.

31.

Griffo, C., Almeida, J. & Guizzardi, G. (2016a). A pattern for the representation of legal relations in a legal core ontology. In Proceedings of 29th International Conference on Legal Knowledge and Information Systems (JURIX), France.

32.

Griffo, C., Almeida, J. & Guizzardi, G. (2016b). Towards a legal core ontology based on alexy’s theory of fundamental rights. In ICAIL Multi-Lingual Workshop on AI and Law (MWAIL 2015), 15th International Conference on Artificial Intelligence and Law (ICAIL 2015).

33.

Griffo, C., Almeida, J. & Guizzardi, G. (2018). Conceptual modeling of legal relations. In ER 2018: Conceptual Modeling (pp. 169–183). doi:10.1007/978-3-030-00847-5_14.

34.

Guerson, J., Almeida, J. & Guizzardi, G. (2014). Support for Domain Constraints in the Validation of Ontologically Well-Founded Conceptual Models (Vol. 175, pp. 302–316). Springer.

35.

Guerson, J., Sales, T.P., Guizzardi, G. & Almeida, J. (2015). OntoUML lightweight editor: A model-based environment to build, evaluate and implement reference ontologies. In 19th IEEE Enterprise Computing Conference (EDOC 2015) – Demonstration Track.

36.

Guizzardi, G. (2005). Ontological foundations for structural conceptual models. PhD thesis, Telematica-Institut / CTIT.

37.

Guizzardi, G. (2006). The role of foundational ontology for conceptual modeling and domain ontology representation. In 7th International Baltic Conference on Databases and Information Systems (pp. 17–25).

38.

Guizzardi, G. (2007). On ontology, ontologies, conceptualizations, modeling languages and (meta)models. In Proceedings of the 2007 Conference on Databases and Information Systems (pp. 18–39).

39.

Guizzardi, G. (2013). Towards ontological foundations for the conceptual modeling of events. In ER 2013 (Vol. 8217, pp. 327–341). Heidelberg: Springer.

40.

Guizzardi, G., Falb, O.R. & Guizzardi, R. (2008a). Grounding software domain ontology in the unified foundational ontology (UFO): The case of ODE software process ontology. In IDEAS ’2008.

41.

Guizzardi, G., Falbo, R. & Guizzardi, R.S. (2008b). Grounding software domain ontologies in the unified foundational ontology (UFO): The case of the ode software process ontology. In 1th Iberoamerican Workshop on Requirements Engineering and Software Environments (IDEAS ’2008).

42.

Guizzardi, G., Guarino, N. & Almeida, J. (2016). Ontological considerations about the representation of events and endurants in business models. In BPM 2016 (Vol. 9850, pp. 20–36). Cham: Springer.

43.

Guizzardi, G. & Wagner, G. (2010). Towards an ontological foundation of discrete event simulation. In Winter Simulation Conference (pp. 652–664).

44.

Guizzardi, G., Wagner, G., Falbo, R. & Guizzardi, R.S. (2015). Towards ontological foundations for conceptual modeling: The unified foundational ontology (UFO) story. Applied Ontology. doi:10.3233/AO-150157.

45.

Hague-Visby (1968). The Hague Visby rules, Brussels protocols.

46.

Hamburg (1978). United Nations Convention on the carriage of goods by sea, 1978 (Hamburg rules).

47.

Hoekstra, R., Breuker, J., Di Bello, M. & Boer, A. (2009). LKIF core: Principled ontology development for the legal domain. In Proceedings of the 2009 Conference on Law, Ontologies and the Semantic Web: Channelling the Legal Information Flood. Frontiers in Artificial Intelligence and Applications (Vol. 188, pp. 21–52). IOS Press.

48.

Kelsen, H. (1991). General Theory of Norms. Oxford: Clarendon Press.

49.

Kelsen, H. (2005). Pure Theory of Law. The Lawbook Exchange. New Jersey: LTD.

50.

Klatt, M. (2012). Robert Alexy’s Philosophy of Law as System. Oxford University Press.

51.

Kowalski, R. (1974). Predicate logic as a programming language. In Proceedings of IFIP Congress (Vol. 74, pp. 569–574).

52.

Mommers, L. (2003). Application of a knowledge-based ontology of the legal domain in collaborative workspaces. In Proceedings of the 9th International Conference on Artificial Intelligence and Law (pp. 70–76). doi:10.1145/1047788.1047799.

53.

Mylopoulos, J., Jureta, I.J. & Faulkner, S. (2007). An ontology for requirements. In Advances in Conceptual Modeling – Foundations and Applications. ER 2007. LNCS (Vol. 4802). Berlin, Heidelberg: Springer.

54.

O’Connor, M., Shankar, R., Nyulas, C. & Das, A. (2008). Developing aWeb-based application using OWL and SWRL. In AAAI Spring Symposium, USA.

55.

Palmirani, M., Ognibene, T. & Cervone, L. (2012). Legal rules, text, and ontologies over time. In RuleML@ECAI 2012.

56.

Presutti, V., Daga, E., Gangemi, A. & Blomqvist, E. (2009). EXtreme design with content ontology design patterns. In Proceedings of WOP 2009 (pp. 83–97). USA: ACM.

57.

Raz, J. (1974). Kelsen’s theory of the basic norm. American Journal of Jurisprudence, 19, 94–111. doi:10.1093/ajj/19.1.94.

58.

Ruy, F., Guizzardi, G., Falbo, R., Reginato, C. & Santos, V. (2017). From reference ontologies to ontology patterns and back. Data and Knowledge Engineering, 109, 41–69. doi:10.1016/j.datak.2017.03.004.

59.

Ruy, F., Reginato, C., Santos, V., Falbo, R. & Guizzardi, G. (2015). Ontology engineering by combining ontology patterns. In ER 2015. LNCS (Vol. 9381, pp. 173–186). Cham: Springer.

60.

Sanchez, D., Batet, M., Martinez, S. & Domingo-Ferrer, J. (2015). Semantic variance: An intuitive measure for ontology accuracy evaluation. Engineering Applications of Artificial Intelligence, 39.

61.

Visser, P.R. & Bench-Capon, T.J. (1997). A comparison of two legal ontologies. In Working Papers of the First International Workshop on Legal Ontologies. Australia: University of Melbourne.

62.

Wyner, A. (2009). An OWL ontology for legal cases with an instantiaiton of Popov v. Hayashi. In Pre-Conference Workshop on Modelling Legal Cases at the 12th International Conference on AI and Law.

63.

Wyner, A. & Hoekstra, R. (2010). A legal case OWL ontology with an instantiation of Popov v. Hayashi. The Knowledge Engineering Review, 14(2), 1–24.