Normative Acts are important legislative and regulatory documents made by different governmental organs. Every year, a huge amount of information is provided in Normative Acts by these organs without control, i.e., there is no effective way to verify redundancies, inconsistencies, crossimpact and ambiguities. In this paper, we propose a domain ontology for Normative Acts based on official documents (the Brazilian Constitution and the Redaction Manual of the Presidency of the Republic) as a reference model that can be used to improve communication, interoperation and automation of Normative Acts. The reference model is built with a highly expressive wellfounded language within a methodology that ensures its quality.
In order to produce the desired impact to society, the NAs must be carefully planned in their elaboration stage, involving studies and researches of previous related NAs. Today, these studies and researches are done manually, without intelligence provided by computational applications – to research a NA, an editor must use a nonspecialized research tool, just like any layman would. Moreover, once published, its content must be easily accessible and researchable by society and other stakeholders. In fact, in Brazil, a federal law ensures that every public organ or entity must publish, in detail, the formats used to structure their information. It is widely known that, although a reference document exists for the NAs writing, this document is not always used by the legislative houses and regulatory agencies, thus resulting in the above cited problems.
The official reference document which deals with the NAs’ writing is the Presidency Writing Manual (in Portuguese, Manual de Redação da Presidência da República – available at http://www.planalto.gov.br/ccivil_03/manual/). The Writing Manual is divided into two parts: the first, among other topics, presents the official communications and standardizes the layout of expedients. The second part deals with the elaboration and wording of NAs and presents examples of the Normative Acts and the legislative procedure.
This second part of the document, in particular, describes the textual elements
that a NA may have and their inter-relationships, making it a valuable reference and a
natural candidate for the development of an Ontology Reference Model for this domain.
Even though the description contained in the Presidency Writing Manual is available
only in natural language (Portuguese), which does not guarantee absence of ambiguities,
the use of an ontologically well-founded language can identify and correct these
possible deficiencies. Such type of language differs from other commonly used
languages to represent data or knowledge in information systems (like databases
schemas) as they are built accordingly to a foundational ontology, i.e., a meta-ontology
that describes a set of real-world categories that can be used to talk about reality
[
Our objective in this paper is to present a domain ontology developed to be a
reference model for NAs. Reference Models are essential artifacts when dealing with
information of a given domain, as they formalize concepts and their relations in a clear
and unambiguous way, improving communication and information exchange and
interoperation. In fact, the main objective of a reference model is to assist humans in
tasks such as meaning negotiation and consensus establishment. This goal can be
achieved by using highly expressive languages, within a formal ontology engineering
methodology, to create a strongly axiomatized ontology that approximates as well as
possible to the domain conceptualization. The focus on these languages is on
representation adequacy, instead of computational representation [
In brief, the main contribution of this paper is to provide an ontology reference model for the NAs domain, formalized with OntoUML within an ontology engineering methodology. A secondary contribution is the presentation of this ontology`s capabilities to be used as basis for computational implementation.
The work described here is placed in the context of the Information and Knowledge Management Model Project (“Modelo de Gestão da Informação e do Conhecimento”, MGIC, in Portuguese), a cooperation project between the Federal Fluminense University and the National Agency of Terrestrial Transportation (ANTT). The MGIC aims to improve efficiency in ANTT’s decision-making. To achieve this goal, the information modeling work is performed on four different fronts. The ontology modeling, one of these fronts, aims to create conceptual reference model ontologies for structuring the agency’s information.
This paper is structured as follows: in section 2, we compare this work with others that are related to NAs; section 3 presents the ontology engineering methodology used to create and guarantee the quality of the proposed NAs ontology; in section 4, we describe the ontology reference model for NAs in details; section 5 presents possibilities of practical use of the NAs ontology as a conceptual model and as basis for computational implementation; section 6 presents the main conclusions as well as future works.
As the legislative process involves several public entities in various spheres and as it
generates a large amount of information, a huge number of works are published
involving the creation of ontologies used for communication and automation in this
domain. In fact, some of these works are the results of large projects, like the LexML
Project [
Legal ontology projects differentiate from each other in scope, objective and in
which kind of ontology they are based. As an example, the Brazilian LexML project
uses as basis an ontology called FRBROO, an ontology in the domain of cultural
heritage; while the Ontojuris Project uses a lightweight ontology (almost a dictionary),
similarly to other initiatives, such as the LTS (Legal Taxonomy Syllabus) [
The Power project, in its turn, uses shared conceptual models to facilitate the
legislation process [
Even though there is a vast number of works about ontologies in the legislative area, it appears, however, that no work involves the creation of a well-founded ontology reference model for the representation of the Normative Acts’ internal structure.
In order to create an ontology reference model that correctly reflects the intended
domain and that is able to be used by different agents (people, groups of people and
other, like machines) to interoperate, an ontology engineering methodology must be
used. This section presents our methodology, partially based on the Ontological
Approach to Domain Engineering presented in [
Scope Definition
Information Acquisition
Syntatic Validation Anti-Pattern Treatment
Conceptualization
Ontology Formalization Simulation
Ontology Reference
Model
Scope definition is the first step of the iterative methodology. Our ontology uses a reference document, the Presidency Writing Manual, written in natural language (Portuguese in our case), as the modeling scope.
The second step of the methodology is the ontology capture, were the sub
activities of Information Acquisition and Conceptualization are realized. In order to
acquire information, a domain study is necessary for the modeler to learn about the
subject to be modeled. We used the Presidency Writing Manual and the Brazilian
Constitution as main source of information. Conceptualizations are immaterial entities
that only exist in the mind of the user or a community of users of a language. In order to
be documented, they must be captured in terms of some concrete artifact. This implies
that a language is necessary for representing them in a concise, complete and
unambiguous way [
The ontology formalization step consists in the formalization, through diagrams,
of the domain model. In order to correctly represent a domain, an expressive language
must be used. This language should be able to represent information despite of
implementation technologies or limitations. In this work, we use OntoUML, an
ontologically well-founded UML profile [
As we intend to create a reference model for Normative Acts (NAs), it is important to ensure that the diagrams allow only instantiation as desired, that is, that the user can only create instances that are possible in the real world. To do this, we focus on the validation of information modeled at the diagrams. In this stage, we have two main types of validation: (1) the syntactic one, which guarantees that the OntoUML models created are syntactically correct, that is, that the entities created are according to the languages meta-model; and (2) the semantic validation, where we want to avoid syntactically correct diagrams that can be instantiated to generate undesired world of affairs.
The OntoUML Lightweight Editor (OLED) (https://code.google.com/p/ontoumllightweight-editor/) provides the syntactical validation. The Semantic Validation does not ensure that there is no impossible state of affairs allowed by the ontology, it does, in fact, ensure that its occurrences are reduced. The semantic validation is done in two steps, within an OLED’s module, called MOVE (OntoUML Model Validation Environment): the first step is an anti-patterns identification and treatment and the second step is a simulation using Alloy.
As stated in [
OntoUML provides a well-founded UML profile. The classes in OntoUML are based
on some important ontological meta-properties that allow the creation of consistent
ontologies [
Examples of meta-properties are identity principle and rigidity. Identity principle is related to the nature of an object. For example, a Student is a Person, as they have the same identity principle, but they can never be a Car, as they have different identity principle. The rigidity principle is the capacity of an entity to be part of a class maintaining its existence. For example: John is an individual of the class Student but, in a given world, it can cease to be a Student and still exists as a Person. However, in any world John cannot cease to be a Person without ceasing to exist. Thus, Student is an example of an anti-rigid class while Person is an example of a rigid class.
The OntoUML stereotypes present in the Normative Acts Ontology are
summarized in Table 1 above. For an in depth presentation, formal characterization and
empirical evidence for a number of the ontological categories underlying OntoUML, the
reader is referred to [
The domain ontology presented here is based on official documents: the Presidency Writing Manual (hereafter, for short, called PWM) and on the Brazilian Federal Constitution. Although the Normative Acts (NAs) Ontology created to the National Agency of Terrestrial Transportation involves three aspects of the Normative Process Structural Elements, Management Issues and Regulatory Marks – the Reference Model presented here considers just the Structural Elements of NAs, considering NAs compositions, aggregations, its internal elements, and all relations between these.
The ontology reference model presented in detail in this section is divided in three subdomains and modeled using OntoUML diagrams. For highlight, the first occurrences of an ontology terms are presented with the Courier New font. It is important to mention that the ontology reference model is fully available for download at: http://www.nemo.inf.ufes.br/en/courses/ontologyengineering.
The model related to this subdomain states the different existing types of Normative Acts in Brazil and their internal structure.
Due to space limitations, the diagram that differentiates the NAs is not presented in image. This diagram’s information was extracted from the 59th Brazilian Federal Constitution article, where it states that the legislative process involves the creation of: Constitution Amendments, Complementary Laws, Ordinary Laws, Delegated Laws, Provisional Measures, Legislative Decrees and Resolutions. Decree, Ordinance and Handout were extracted from the PWM. All of these NAs are disjoint from each other, i.e., no NA can be of two different types at the same time. NAs are defined by its composition by different subkinds of Articles and by its preliminary elements.
At the adopted conceptualization, the Article performs the most important function in NAs as it contains the statements defining the rules and information that the NA is about. In the ontology reference model every concept has a (direct or indirect) relation with Articles. It can be seen in Figure 2 that every article has an identifier number. An OCL rule formalizes that Articles in the same NA have unique natural numbers.
Preamble, Epigraph and Brief (in Portuguese: Preâmbulo, Epígrafe,
and Ementa, respectively) are obligatory preliminary elements in every NA. These
preliminary elements cannot be modified (vetoed, revoked or altered) – these properties
are stated in the Ontology Reference Model by the composition meta-properties of
essential and inseparable [
Grouping and Discrimination Elements are important part of NAs as they provide to their author the desired abstraction granularity. The Discrimination Elements are used to describe in more detail the information being normalized in a NA.
The different types of Discrimination Elements are: Paragraph, Item, Letter (in Portuguese: Parágrafo, Inciso, and Alínea, respectively) and Letter Discriminator. Every Discrimination Element is a part of a Normative Act because the Articles that are discriminated by these elements are part of the Normative Act.
Articles can be of two types: Simple Article or Composed Article. While the former consists only of a text, the latter consists of its introductory text, named Caput, and of at least one Item or Paragraph. Both types are represented in Figure 3.
If there is just one Paragraph composing an Article, its identification string must be Unique Paragraph. OCL rules ensure these restrictions.
Articles and Paragraphs can be decomposed in Items (see Figure 3), identified by roman numbers. Items, represented in Figure 4, can be Simple Items (when undivided) or composed by Letters. Letters can also be undivided (Simple Letters) or they can be divided in Letter Discriminators. The PWM states at its section 10.2.2.3 that “Letters can be discriminated with cardinal numbers, followed by periods”. As no name is given to these discrimination elements, we named them Letter Discriminators.
The Grouping Elements are used to easily aggregate related information in a NA. They are the Parts, Books, Charters, Chapters and Sections (in Portuguese: Parte, Livro, Título, Capítulo, and Seção, respectively). Every Grouping Element is part of the NA that is composed of the Articles that are grouped by these Grouping Elements.
As represented in Figure 5, related Articles can be grouped in Sections or Chapters. Sections can be of two types: Simple Sections, i.e., Sections that are not composed by other Sections, and Composed Sections, that are Sections composed by other Sections or Subsections: a specific type of Section that just occurs composing a Composed Section.
The grouping of Chapters by Charters, of Charters by Books, and of Books by Parts can be seen in Figure 6.
Parts can be of types Double Part or Multiple Part. OCL rules were created to state that every NA that is composed by a Multiple Part always have at least three of these and to state that if the NA is composed of a Double Part it must have exactly only one instance of a General Part and a Special Part.
As stated by [
In Brazil, according to the Federal Information Access Law (Brazilian Federal Law number 12,527, from November 18th 2011. Available just in Portuguese at http://www.planalto.gov.br/ccivil_03/_ato2011-2014/2011/lei/l12527.htm), every public organ or entity must use the Internet to publish, in detail, the formats used to structure their information. The ontology presented here, as a well-founded formalization of a domain conceptualization, is the best option to accomplish this requirement. This same law also states that the public organ or entity must enable automated access by external systems to the information in open, structured and machine-readable formats. Thus, an ontology implementation in the Resource Description Framework (RDF) or OWL can entirely satisfy the law. In fact, the federal government with the World Wide Web Consortium (W3C), cite OWL as a desirable practice for Open Data [Comitê Gestor da Internet no Brasil 2011].
A computational implementation of the proposed ontology reference model can also be used in applications to assist different stages of the regulatory and legislative process, like the NA edition. The usage of automation software can significantly improve the time spent with the huge amount of NAs that are created daily in a public entity. This automation can make processes more effective as it reduces the domain specialists’ writing and research time and it gives them more time to think and reason about the subject of the documents.
In the edition phase, ontology-based software can eliminate from the writer the need to know each concepts from the PWM, as the software can automatically create these concepts for him. Another great improvement is the restriction to create documents that are not in accordance with the PWM. As an example, every time a writer creates an Article, if he wants to create a subtopic of this Article, he can just press the TAB key and two options appears to him: an Item or a Paragraph. So, it is not possible to create, for example, a Letter discriminating directly an Article. Another possible usage of this software would be to evaluate Normative Acts already created.
The ontology-based software for the NA process can be significantly improved by using other Ontology Models. The usage of an ontology model about NAs’ revocation or modifications can add the ability to control the impact of a newly published NA in others already published. That is, there will be no need to manually state that an NA is repealed; it can be done automatically by software – thus, eliminating human errors.
Also, adding ontology reference models to the domain that the NAs are about,
we can create a semantic annotation feature [
With the usage of semantic annotation, the search on documents is improved, as there is no need to lexically match all search terms. A search using the word “highway” can return documents that use the term highway, and documents that deal with highways but do no explicitly use this concept, for example, a document that contains information about the BR-101. An improved search is vital to guarantee that new NAs do not conflicts with older NAs.
Already existent ontology-based tools for the legislative process can be found in
[
This work presented an Ontology Reference Model for the domain of Normative Acts, built over information extracted from the Brazilian Constitution and the Presidency Writing Manual. The proposed ontology is formalized using an ontologically wellfounded highly expressive language and it was developed following the ontology engineering methodology presented here. This ontology can be used as basis for automation and interoperation of information about this domain.
The Ontology Reference Model is presented considering three subdomains: (a) Normative Acts and Articles, where different existing types of Brazilian Normative Acts are presented as well as their internal structure, representing their composition by articles and preliminary elements; (b) Discrimination of Normative Acts’ Elements, related to elements that are used to describe in detail the information being normalized in a Normative Act; (c) Grouping of Normative Acts’ Elements, that formalizes elements that are used to easily aggregate related information.
As a reference model, this ontology aims to make a clear and precise description of the domain elements for the purposes of communication, learning and problem solving. As an implementation, the model is able to solve problems related to the redaction and edition of Normative Acts, like reference finding, and cross-impact analysis. The ontology, in its conceptual or computational form, can be used to adequate public organs to the Federal Information Access Law, as every public organ or entity must use the Internet to publish, in detail, the formats used to structure their information.
Future work involves the expansion of the ontology to represent other aspects of Normative Acts, like temporal ones (e.g. an NA is valid from a beginning date and can have an end date). Another important future work is the creation of the computational ontology in semantic web languages, like OWL.
Acknowledgements. This research has been funded by the MGIC Project. We thank the ANTT employees and all project members who have somehow contributed to the development of this work. This research has also been funded by FAPES/CNPq (PRONEX 52272362/11).