Semantic Interoperability for Legal Information: Mapping the European Legislation Identifier (ELI) and Akoma Ntoso (AKN) Ontologies

The legislative landscape, characterized by overwhelming amounts of legal data which, on many occasions, is only accessible by legal experts, the fragmented nature of information and the ever-increasing number of disparate systems, have given more impetus to the interoperability realm of legal data. The semantic interoperability of Linked Open Legal Data (LOLD) requires rich and well-defined metadata, as well as the establishment of standards, in order to be able to connect and link these scattered legal data resources. This is usually achieved through the transformation of legal information into structured format and through the utilization of legal ontologies whose main purpose is to connect the legal basis of two or more countries by allowing for reusability and the ability to adequately represent legal information, which is understood and retrieved across borders. Within the context of this study, the European Legislation Identifier (ELI) and Akoma Ntoso (AKN) ontologies are mapped in order to make legal data compatible and reusable in as many contexts as possible and to support the Linked Open Legal Data (LOLD) concept. The mapping of these two widely used legal ontologies was evaluated by domain experts and strongly validated by tools. The usefulness of the produced mapping is proven through its real-life context application, although one thing to consider regarding possible future perspectives, could be the inclusion and mapping of more legal ontologies to expand the application domain and improve the semantic interoperability of legal information. These mappings could be either achieved using similar methodological approaches or applications of automated and AI-based ontology mapping techniques.


INTRODUCTION
Legal Informatics refers to the discipline which deals with the application of Information and Communication Technologies (ICT) to the processing of legal information, and the provision of support to legal activities within the context of the legal environment [1][2] Legal information is usually openly accessible by default, but on many occassions (if not most) in a very restricted and nonmachine-processable format, such as plain document files.This can make it very difficult for the interested user to identify the piece of information they are looking for.Users are often confronted with the challenge of accessing vast, hitherto untapped, sources of legal information in an efficient manner that not only provides them with all relevant information, but also assists and supports them substantially in day-to-day tasks [2].Furthermore, recent trends in digitization, open data, and social media have resulted in an exponential increase in the amount of data available for use by policy makers in order to make sense of the socio-economic and political phenomena, and design relevant public policies [3].Legal informatics is a field of practice expected to provide multiple advantages for the community in general, as well as the legislative procedures and policy making.To support the provision and use of legal data and to achieve the realization of the aforementioned gains, an increasing number of legal data artefacts and portals, both proprietary and open, have been designed and developed.
As the number of ontologies that are made publicly available and accessible on the Web increases steadily, so does the need for applications to use them [4].A single ontology is no longer enough to support the tasks envisaged by a distributed environment like the Semantic Web, thus multiple ontologies need to be accessed from several applications.Ontology mapping could provide a common layer from which several ontologies could be accessed and hence could exchange information in semantically sound manners [4].Put differently, the vastness of the increasing ontological space, has increased the need for interconnection among the ontologies in order to allow for their compatibility, and reusability, to the optimal extent, whenever this is applicable.
Over the last decade the legal informatics community has devoted large effort to define technical standards for modelling legal documents (Akoma Ntoso [12], LegalRuleML [9], ELI [8], [10], ECLI [7], and more) with the aim of supporting open access, reusability, interoperability, and findability of legal knowledge on the Internet.Several official gazettes dematerialized their publications (AKN4EU, AKN4UN, Luxembourg, The National Archives UK, Library of Congress of Chile, The US Code, GPO US, etc.), and Parliaments (Senate of Italy, European Parliament, European Commission, UN, etc.) started to use XML techniques for the annotation of legislative sources to enhance the internal workflow and to improve the lawmaking process from the beginning.In this scenario, the courts and tribunals, and the judiciary system, in general, are one step behind in the standardization of legal knowledge and documentation.
The purpose of the presented study is to examine how it is possible to utilize the European Legislation Identifier (ELI), to reuse ECLI URI and metadata, and to map them with a unique international standard like Akoma Ntoso (AKN) that is able to i) manage both legislative and judiciary documents in a unique XML format; ii) include different ontological and naming convention levels; iii) create a distributed database among different collection of legal documents; iv) implement navigation of the references; v) design advanced searching, visualization and legal data analytics applications.This examination is being done in the framework of the Manylaws1 INEA-CEF EU project which developed an interoperable infrastructure for EU, Austrian and Greek legislation systems.The intended outcome of this research is to contribute to the semantic interoperability in the legal domain by mapping the ELI and AKN ontologies, but also considering their compatibility with the DCAT [11] standard (which, consequently, also ensures compatibility with the EU data portal), in order to make them compatible and reusable in as many contexts as possible and to support the Linked Open Legal Data (LOLD) concept.
The article is organized as follows: in the Background section, a view of the current status on legal informatics and ontologies and an overview of various approaches in ontology mapping, as well as the descriptions of the ELI and AKN ontologies will be presented, while the Research Method section contains the description of the proposed methodological approach.The Results section consists of the mapping between the two ontologies, while the Discussion section includes the interpretation of the findings in a meaningful context, as well as potential limitations of the presented study and ways this research could be expanded in the future.

BACKGROUND
The legislative landscape, characterized by overwhelming amounts of data, the legal information which, on many occasions, is only comprehensible and accessible by legal experts, the fragmented nature of information and the ever-increasing number of disparate systems, have given more impetus to the interoperability realm of legal data.The semantic interoperability of Linked Open Legal Data (LOLD) requires rich and well-defined metadata, as well as the establishment of standards, in order to be able to connect and link the scattered legal data resources.This is usually achieved through the transformation of legal information (which could often be plain text or document) into structured format, and this is where the role of legal ontologies kicks in.One of the main purposes of legal ontologies is to manage to connect the legal basis or framework of two or more countries by allowing for reusability among different systems and the ability to adequately represent legal information, which is understood and retrieved across borders [5].In this vein, various legal ontologies have been developed, among them being the European Legislation Identifier (ELI), the Akoma Ntoso (AKN), the Legal Knowledge Intechange Format (LKIF), the CEN Metalex, the European Case Law Identifier (ECLI), and the LegalRuleML.Loutsaris & Charalabidis (2020) [6], presented an overview of those ontologies, as well as their utilization by Legal Information Systems Initiatives.
The quest for Big Linked Open Legal Data (BLOLD) has led various European projects to adopt such ontologies, Manylaws, Lynx, EUCases, Openlaws.eu,Caselex, and N-Lex being some of them.These initiatives adopt legal ontologies, such as the aforementioned ones, in order to achieve their intended purpose.As Loutsaris & Charalabidis (2020) [5] showcase in their study, the common denominators for such initiatives, especially when their intended outcome is multinational and not restricted to the border of solely one country, are the AKN and ELI ontologies.AKN and ELI are widely used in the legal domain, which gives them a spot in the priority list for making them reusable to increase the compatibility and interoperability of legal information.The present research takes into consideration the described situation and, thus, the mapping of ELI and AKN, as well as the compatibility with DCAT [11] are entering the scope of the study.
I. The European Legislation Identifier (ELI) "The European Legislation Identifier (ELI) is a system to make legislation available online in a standardised format, so that it can be accessed, exchanged and reused across borders." (Eur-lex.europa.eu,2023).The core ELI ontology specifies a data model to structure the data into machine-processable format, it includes identifiers (URIs) of information and metadata specifications for information description.ELI is mainly dealing with the description of legal documents and their basic metadata schema.The backbone of the ontology is shown in Figure 1 (op.europa.eu,2023).
The legal documents are expressed according to the aforementioned levels in order to achieve a standardized description of information, compatibility among the EU countries in agreement, and metadata serialization.
II. Akoma Ntoso (AKN) "Akoma Ntoso ("linked hearts" in the Akan language of West Africa) defines a set of simple technology-neutral electronic representations in XML format of parliamentary, legislative and judiciary documents" [12].As a point of difference with ELI, the purpose of AKN is not solely the expression of legal documents, but it also breaks down the structure of the legal document in the articles, paragraphs, and other components it consists of.It is also linked to parliamentary data.For this reason, a potential mapping between AKN and ELI needs to be performed by identifying the element categories of AKN which can be adapted and reused in the ELI, thus transferring the capabilities of this ontology to the other.
Ontology Mapping Methods In general, ontology mapping can be achieved in a wide variety of ways, in the spotlight being various text mining or other AI-based techniques.Lv & Peng (2020) [13] applied evolutionary algorithms (EAs), and more specifically, the grasshopper optimization algorithm in order to improve the optimization process of aligning (or mapping) two ontologies.This process involves the use of similarity measures to show the score of proximity between concepts included in the ontologies under examination to help with the mapping of highly similar and compatible elements.Chong & Lee (2022) [14] applied a deep learning -based ontology alignment approach, which targetted heterogeneous data in different domains and aimed to resolve semantic interoperability issues by combining unsupervised (for the representation of ontologies) and supervised (for further enhancement) learning.On another note, Oliveira & Pesquita (2018) [15] performed a pattern analysis of existing biomedical ontologies in order to explore the potential challenges in aligning them, while  2015) [21].Moreover, the various categories of ontology mapping (the mapping between an integrated global ontology and local ontologies, the mapping between local ontologies, and the mapping in ontology merge and alignment), as well as the pitfalls and pros of the aforementioned, were described by Choi et al. (2006) [22] as early as in 2006.
Looking at the ontology mapping landscape for legal information, specifically, there appears to be very limited (if not almost non-existent) literature describing ontology mapping implementations for legal information, in specific.In this light, the authors aim to contribute to this lack of related literature by proposing a methodology for ontology mapping in the legal domain.

METHODOLOGY
At this point, the adopted methodological approach for the presented study is explained and described.The method applied in this case is a combination of manual ontology mapping and qualitative analysis (experts' opinion), while the final artefact evaluation is supported by tool validation techniques.An overview chart of the applied methodology and the phases it consists of, is shown in Figure 2.This flowchart shows the procedure to finalize the ontology mapping of ELI and AKN starting from the initial mapping performed by the authors, and followed by the experts' opinion and feedback loop until consensus is reached for all the mappings presented.
To begin with, the authors examined the description of the Akoma Ntoso schema and its main properties and relations (Name, Description, Based on, Applicable to etc.) and then identified the equivalences of the same in the ELI.What was especially catered for, was to ensure the compatibility with the DCAT standard model The mapping was initially conducted in a manual fashion in order to gain the first overview of the mapping by considering the common elements, differences in annotation and taking into account the description for each to avoid misinterpretations.
As a next step, the initial mapping was presented to a number of domain experts (experts in legal information representation) and feedback was requested by them.Most of the experts (4) are parliamentary specialists in two European Union countries and 1 expert is a legal informatics professor.The 5 domain experts involved in the procedure went through the initial mapping and provided their comments and feedback by annotating each mapping with a color, according to their confidence about the mapping.So, the mappings they completely agree with are denoted in green color, the mappings they believe are partially sufficient are denoted in yellow color, while the mappings they do not agree with are denoted in red color.
An example of the followed process is presented in Table 1.Here, an excerpt showing the equivalences between ELI and AKN is presented.In this example, one of the interviewed experts has indicated in green color the mappings he/she has agreed upon ("repeal" of AKN to "repeals" of ELI, "join" to "consolidates" and "renumbering" to "correct").The mappings he/she partially agreed on is indicated in yellow color ("substitution" to "amends" and "insertion" to "changes") and these are the mappings which, as demonstrated in the methodology flowchart of Figure 2, are the ones taken to the Workshop phase (along with the red ones) in order to be discussed again until a full agreement of all 5 experts is reached for the final mapping.
After the experts provided their feedback according to the aforementioned process, the collected information was then prepared in the form of follow-up plans on discussion, which was held with the experts during an extensive Workshop session.During the Workshop session, the authors and the interviewed experts went through all the mappings denoted in yellow and red in order to discuss and propose the final mapping.At this point, it is important to mention that, as far as the mappings denoted as "green" are concerned, in order for them not to be included in the Workshop discussion, a consensus needed to have been met among the experts.So, if a mapping was considered green by 3 experts but yellow by the other 2, then it was still brought up in the discussion.The process continued in this manner until all the mappings were evaluated by the experts during the organized Workshop session, and until a consensus was achieved among all the experts.It is worth mentioning that it was a very long and difficult process to come up with consensus among all experts on such technical issues.The discussions on "red flags" followed the way of supporting the covered by Manylaws project legislation systems.Another level of validation of the final mapping was provided by the checkers of DCAT and AKN and they are included in the Results section.

RESULTS
Having followed the steps described earlier in the Methodology section, the results of the mapping are presented in Table 2.The mapping of Table 2 is the final outcome of the mapping having taken into consideration the experts' feedback provided during the Workshop discussion to reach complete consensus on the results.The mapping was performed on the ELI ontology version 1.2 with some evolutions and modifications from 1.1 and on the AKN version 1.0.
As mentioned previously, Table 2 shows an excerpt of the final mapping of ELI to AKN.The columns include the principles and the respective description for each ontology, followed by the last column (Remarks) which includes information about the conducted alignment while also including potentially useful commentary for the judgement of the performed mapping.For the sake of briefness, Table 2 is indicative and it presents a big part of the mapping but it is not exhaustive.

DISCUSSION
Standardization is a typical and important step in the technology lifecycle, affecting both innovation and diffusion, and generally indicating an advanced stage of technology development.In the domain of legal informatics, standardization has taken place, not only in the sense of informal standards emerging but also in the sense of the formal promulgation of a standard by a standardizing body.
Since metadata is a key resource to semantic technologies, when it comes to improving the access to legal and legislative documents manually edited as well as machine generated metadata, their role becomes primary and ways to improve metadata interoperability need to be examined.This is the reason metadata conventions and metadata standards demand particular attention.At this point, the importance of the compatibility of standards and ontologies can be inferred, making the research presented in this study highly relevant.The ontology mapping presented can be used in various contexts of legal information representation and handling, one of them being the "Manylaws" project, for instance.The "ManyLaws" project, funded by the Connecting Europe Facility, intended to build a portal that, within the framework of the European Data Portal (EDP), can provide trans jurisdictional and translingual access to legal and legislative documents, whose metadata are going to be harvested as well as enriched by AI based algorithms applied on the documents within a high-performance computing (HPC) environment.Thus, the "Manylaws" metadata convention needed to be The publication is a part of the meta block and captures the metadata concerning the official publication process.The name of the source (Official Journal), the date (in normal format), the label for the presentation (showAs), the number of the source of publication (number).The publication is mandatory for the act type but not for the bill type of document.
The AKN field of publication contains all the information about the publication including the document number which is a different element in ELI.

Publication name
The name of the source (Official Journal) In ELI the publication name is always the Official Journal.date the date (in normal format) date_publication Date of publication of the official version of the legislation, in hard copy or online, depending on what the official publication is, and when it was published.
Publication dates at the level of legal expressions can be separately asserted, using standard Dublin Core properties.
In the Manylaws schema we are keeping information about both AKN and ELI fields since these required fields for validation in both ontologies.

lifecycle
The lifecycle lists all the events that are involved within the chain of modifications of the document.These events modify the expression.
In on a voluntary agreement between the EU countries.It includes technical specifications on web identifiers (URIs) for legal information, metadata specifying how to describe legal information, and a specific language for exchanging legislation in machine-readable formats.• Akoma Ntoso which has been standardized by the Legal-DocML Technical Committee within OASIS.As mentioned earlier, while the ELI metadata schema is based on a comparative analysis of the legal documents of the jurisdictions of the EU and its member states, Akoma Ntoso has its basis in a world-scale comparison of legislative, legal, and judicial documents; thus, it is more comprehensive than ELI, and has a particular focus on the life cycle of legal documents.
Given the variety as well as the relevance of legislative and legal documents, mapping the standards created an improved framework for describing them, thus, making it of high importance.In this scenario, the judiciary system, in general, is one step behind in the standardization of legal knowledge and documentation.The "Manylaws" project extracts the necessary information of a law using text mining techniques and then annotates the legal information in a structured format [6].
Although the usefulness of the produced mapping is proven through its real-life context application, some limitations are also present.One of the most important things to consider is that, as mentioned earlier, the presented study involves the mapping of the most basic and commonly used ontologies in the legal domain, but not including all of them exhaustively.Understandably, this might not need to be the case that all of the existing legal ontologies necessarily ought to be mapped, however, the examination and inclusion of more ontologies until a defined level of semantic interoperability is achieved, is most probably a necessity, considering the rapidly evolving information landscape.
In this light, future considerations with respect to this research could be the inclusion and mapping of more legal ontologies to support and improve the semantic interoperability of legal information and its compatibility with as many standards as possible.These mappings could be either achieved using similar methodological approaches, such as the one presented in this study, applications of automated ontology mapping based on AI or similar technologies, as described in the Background section, or a combination of various methods, taking the best out of machine and human expertise for cross validated results.

CONCLUSIONS
The intended outcome of this research has been to enhance semantic interoperability in the legal domain by mapping the ELI and AKN ontologies, but also considering their compatibility with the DCAT standard (thus ensuring compatibility with the EU data portal), in order to make them as highly compatible as possible, make them reusable in various contexts and support the Linked Open Legal Data (LOLD) concept.This is usually achieved through the transformation of legal information into structured format and through the utilization of legal ontologies whose main purpose is to connect the legal basis of two or more countries by allowing for reusability and the ability to adequately represent legal information.Within the context of this study, the European Legislation Identifier (ELI) and Akoma Ntoso (AKN) ontologies were mapped following a specified methodology including manual mapping and experts' opinions in an iterative process until the final mapping was established and validated.The mapping of these two widely used legal ontologies was evaluated by the interviewed domain experts and strongly validated by the tools.The usefulness of the produced mapping is proven through its real-life context application, as presented in the "Manylaws" project instance and the related literature, which clearly states the need for the unification of at least the most dominant ontologies per domain to allow for compatibility and standardization, but also to alleviate the increasing problem of disparate data resources and heterogeneous information systems.The contribution of this work is, first of all, to propose a methodological approach which could be applied in various contexts and combined with automated, semi-automated and human-expertise focused approaches for ontology mapping, but also, to start paving the way towards the mapping of basic ontologies in a particularly demanding domain, such as the legislative one, where relevant progress with respect to semantic interoperability still appears to be in its infantry.

Figure 1 :
Figure 1: The backbone of the ELI ontology (op.europa.eu,2023) Stavropoulou et al. (2020) [16] designed and developed a big open legal data analytics platform based on text mining, semantic analysis and interactive visualization methods.Kiourtis et al. (2019) [17] developed a methodology based on Structure Mapping in order to transform collected healthcare data into Health Level 7 Healthcare Interoperability Resources (FHIR) format to achieve interoperability in healthcare.Apart from the various technologies supporting innovative ways to map ontologies, a number of ways to evaluate and compare ontology alignment systems have been introduced, for instance, the methodology proposed by Mohammadi & Rezaei (2020) [18] which was based on multi-criteria decision-making (MCDM).A summary of the most recent advancements in ontology mapping and proposed solutions to improve this process was provided by Harrow et al. (2019) [19], while theoretical ontology mapping optimization and other techniques and approaches were described by Chu et al. (2020) [20], and Ardjani et al. (

Figure 2 :
Figure 2: A flowchart of the applied methodology for the ELI and AKN mapping

Table 1 :
An instance of the experts' evaluation process on the initial mapping

Table 2 :
An excerpt of the final mapping of the ELI and AKN ontologies