Methodology - martynui/OGDL4M GitHub Wiki

The Methodology named MeLOn (Methodology for building Legal Ontology) was used to develop the OGDL4M ontology. The MeLOn is a new empirical methodology for building legal ontologies developed by M. Palmirani in order to help legal experts model legal concepts using the principles of data modelisation. MeLOn has already been implemented by a few scholars, and it takes its inspiration from SAMOD.

MeLOn was developed after several years of empirical practice in CIRSFID, and it aims to resolve typical issues working in the legal domain: 1) Legal experts: they lack competencies in conceptual or data modelling, and they often adopt technical tools (e.g., Protégé) without the necessary awareness of the technical consequences; 2) Legal domain sources: legal texts and other relevant sources (e.g., soft law, case law, interpretation, doctrines, social rules) are the main sources for developing a legal ontology, and it is essential to connect existing legal material (whether formalised or not) to the ontology; 3) Legal domain goals: ontologies are often designed teleologically from the start by formalising the goals to be addressed, although in the legal domain we are not limited to one particular application; rather, our aim is to model existing legal concepts “as is”; 4) Legal domain evaluation: evaluation is fundamental for testing the quality of an ontology, but it can be very difficult to evaluate legal concepts. There are problems of exceptions and interpretation, and special methodology should be defined for those use-cases.

MeLOn describes ten steps for creation of an ontology: i) description of the ontology goals and proposing in natural language the definition of some use-cases for the empirical test; ii) definition of evaluation indicators; iii) analysis of the state of the art for related ontologies; iv) formation of a list of all the relevant terminology and production of a glossary of the main legal concepts; v) modelling a knowledge base of the legal domain by creating the following tables (Concepts tables, Object properties, Data properties, Ontology restriction (Axioms)); vi) transforming the tables in UML and later in OWL in order to optimise the modelisation; vii) empirical testing of some scenarios and use-cases defined in step i); viii) refinement of the ontology based on the results of the empirical test, including the evaluation of legal experts; ix) evaluation on the basis of the indicators defined in step ii); and x) publishing and documentation (using LODE tool).

Description of the ontology is a short description of the goal of the ontology in one page with the research questions that the ontology intends to address. Two or three use-cases are selected and described in details (storytelling). In our case the use-cases are Creative Commons licenses, Open Government License Canada v.2.0 and UK OGL v.3.0.
Definition of evaluation indicators defines some parameters/indicators for evaluating the ontology according to the intended end goal. In our case the indicators are the following: i) completeness of the legal concepts definition; ii) correctness of the explicit relationships between legal concepts; iii) coherence of the legal concepts modelisation; iv) applicability to concrete use-case; v) effectiveness for the goals; vi) intuitiveness for the non-legal experts; vii) computational soundness of the logic and reasoning; viii) reusability of the ontology and mapping with other similar ontologies.
State of the art of related ontologies describes the state of the art of related ontologies and answers these questions: Does any ontology already exist that can help to develop the new ontology? If there are any ontologies that can help to develop the new ontology, can the existing ontology be extended or linked to the new one? In our case several existing ontologies were taken in consideration and reused (e.g., L4LOD, LKIF, CopyrightOntology, Time, DBO).
Formation of a list of all the relevant terminology and production of glossary is the process used to develop a knowledge base of the specific legal terminology relevant for the domain and to generate the glossaries. Legislative documents, case law and other sets of legal norms should be consulted for determining the specific legal terminology. A glossary of terminology has the form of a table with these column headings: term, definition by legal source (citing legal source, license, document, case law or legal theory, or common custom of the legal domain), link to normative/legal source, normalised definition (definition of term, made by the author of the new ontology, simplified or extended from a normative/legal source to fulfill the expectations of possible methodology users). The normalised definition should be a natural language description of the legal text using subject, predicate, object, with the aim to reuse the terms of the glossary as much as possible and avoid duplicative or ambiguous terminology. In this way a legal expert is forced to create triples that can be aggregated later on into more abstract assertions (TBox or ABox). Table 1 presents a representative part of a glossary, representing classes and properties coming from a legal source.
Modelling the knowledge base of a legal domain, for example by creating tables of classes and objects and defining their properties and relationships with other classes and objects.
UML and OWL modelling is a process dedicated to modelling the ontology in OWL. These tools are recommended for use: Protégé or yED with Grafoo extension. We used yED with Grafoo and had developed a UML shcema, but later we used Protégé to develop OWL file.
The Test step is dedicated to testing the ontology. The test is divided in two steps: first the authors model the Creative Commons licenses, Open Government License Canada v.2.0, UK OGL v.3.0 that are the use-cases chosen for the preliminary empirical testing. Secondly, using the LIME editor we annotate and connect the OGDL4M classes to the texts. Thirdly, we test the ontology by selected parameters.
Refinement is a process to refine the ontology with the inputs from the Test step.
Evaluation is a step to evaluate the ontology using the previously described indicators.
Publishing and documentation is the concluding process, dedicated to documentation of the ontology with a tool called LODE and publication of the ontology and connection with other ontologies.

At present, we are at the stage of waiting for technical optimization.