In this paper we present an overview of the PhD thesis, which aims to show the feasibility of a legal knowledge modeling based on the definitions included into legal texts using Semantic Web techniques in order to check compliance of tender documents to EU Directives. We show a hybrid approach, i.e. a theory and a system that combine the use of Akoma-Ntoso standard to describe legal texts, OWL 2.0 for modeling legal concepts and the emerging LegalRuleML standard for providing a rule-based extension of legal knowledge representation on the top of ontologies.
The world of public procurements is a field of great significance for many reasons. It is one of the main economic activities of Governments, since through procurements are acquired services and goods for functioning of the res publica and for meeting needs of citizens' communities. This activity represents also a significant business opportunity for economic operators because public procurement in the EU accounts for 17% of EU GDP. In this scenario, the regular conduct of tendering processes plays a key role, in terms of transparency and fairness in bidding competitions, reduction of litigation-related costs, but also corruption fighting strategies.
A tender situation can be considered as a complex case, in which agents, documents
and processes play an interactive role. Norms on public procurements contracts
regulate either the content of documents and the conduct of processes. Through the
representation of knowledge contained in normative texts it is possible to define the
abstract, lawfully type of those documents or processes. Each single document or
applied procedure is an instance of the abstract type defined by normative provisions.
With respect to documents involved in processes related to European public
procurements, both the abstract type and the actual token can be represented with
languages of Semantic Web, in order to describe all the layers perspective of this kind
of texts: structure (as organization of the texts), metadata (as any information that was
not approved by the authority in the document), ontology (as any legal concept called
from text that need a modeling), and finally legal knowledge representation (as
modeling of rules). Through this articulate definition of legal knowledge, as showed
in [
We define compliance the conformity of a legal document and/or of a relevant behavior with the normative prescriptions regulating them. The difference between the prescription stated in legal texts and its actual application is a compliance failure. Through compliance check is possible to detect this failure.
The ultimate goal of the research is to produce a set of pilot cases in order to show the feasibility of such kind of system that would facilitate the monitoring by the administrations on the regular conduct of procedures for the award of public contracts.
XML-based standards are widely adopted by Governments of several countries for
structuring legal documents. Some Legal XML standard (such as Akoma-Ntoso
standard [
The second type of legal knowledge is the dynamic one. The ontological level is used to classify instances (facts) into ontological classes (abstract types). Such classified facts are then subject to certain provisions. This is the point where ontologies and rules meet. Indeed, even if in many cases rules applied to classified instances may be represented through first order logic (necessary condition), in many others cases monotonic reasoning based on Open World Assumptions (OWA) is not suitable for reasoning with knowledge-base inconsistencies or conflicts among rules. In legal domain, accordingly to a set of information an argument can be considered true, but if the set under consideration is enriched, maybe the truth of the argument has to be revised. This type of situation is well known as defeasibility and typically is required non-monotonic logic with Closed World Assumptions (CWA) to represent it. So, our hypothesis is based on the assumption that it is possible to recognize rules that define ontologically legal concepts (in terms of proof-of-concepts) and rules that are applied to legal concepts.
Over the last decades, many rules interchange languages has been produced. The
most relevant are undoubtedly RuleML [
Compliance checking is a field that is typically related to business domain. There is
a vast literature on various techniques developed for supporting compliance checking
of business processes to regulations. Just to name a few, Governatori et al. have been
proposed approaches based on multi-modal logic [
About the legal domain, under the NEURONA project, has been developed a legal
ontology for the representation of data protection knowledge for reasoning about the
correctness of the information regarding personal data files and the correctness of the
measures of protection applied to these data files [
A pilot case scenario is used for explaining the described methodology. It is based on the modeling of a complex norm extracted from the European Directive 2004/18 on the coordination of procedures for the award of public works contracts, public supply contracts and public service contracts, namely the Art. 17. With this example, we intend to show what are both the limits and the role of ontologies versus rules and also that LegalRuleML allows to model articulated types of rules, which may be encountered in modeling of legal knowledge.
The article 17 of Directive 2004/18 states: “Without prejudice to the application of Article 3, this Directive shall not apply to service concessions as defined in Article 1(4)”. In this norm are contained: a reference to a the legal concept of service concession as defined by the same directive (in Art. 1 (4)); an exception to the application of the Directive (“shall not applied”) and an exception of exception about this rule in all the cases covered by the Art. 3 recalled by the Art. 17 (“Without prejudice to the application of Article 3”). In legal domain this mechanism (to recall another rule) is called meta-rule, since it is a rule about (the activation of) another rule, namely the Art. 3. This article states that if a contracting authority grants special or exclusive rights to carry out a public service activity to an entity other than such a contracting authority, the act by which that right is granted shall provide that, in respect of the supply contracts which it awards to third parties as part of its activities, the entity concerned must comply with the principle of non-discrimination on the basis of nationality1. Section 4.1 describes our proposal about modeling of legal concepts (static knowledge); Section 4.2 describes how it is possible to model complex metadata on the top of legal concepts with a syntax look-like LegalRuleML2.
As outlined in Section 1, legal concepts are basically identified by terms incorporating inferential links between preconditions and legal consequences, in the same way of rules. So, the role to express this kind of rules can be delegated to ontologies and in that way the proof of legal concepts can serve as basis of proof of legal compliance, since every individual belonging to a class of a specific legal concept have to comply with the restrictions on this class. For the purpose of the presented pilot case, we've developed an ontology3 for describing particularly the class of Service Concession as defined in the Art. 1 (4) of the Directive 1 Article 3. Granting of special or exclusive rights: non-discrimination clause - Where a contracting authority grants special or exclusive rights to carry out a public service activity to an entity other than such a contracting authority, the act by which that right is granted shall provide that, in respect of the supply contracts which it awards to third parties as part of its activities, the entity concerned must comply with the principle of non-discrimination on the basis of nationality. 2 Documents of the OASIS LegalRuleML TC are available at https://www.oasisopen.org/committees/documents.php?wg_abbrev=legalruleml. The mailing list describing the work in progress can be browsed at https://lists.oasis-open.org/archives/legalruleml/. 3 Available at http://codexml.cirsfid.unibo.it 2004/18/ec and the class of Act of grant to entities others than authority as defined in the Art. 3 of the Directive 2004/18/ec. The main commitments of this ontology are to check if an individual of type Public Procurement
Contract is also of type Service Concession, and which individual belongs to the class of Service Concession, but is also of type Act of grant to entities others than authority. In that way the rules are applied upon the materializations made by the ontology on legal concepts. According to the definition of Art. 1 (4) of the Directive 2004/18/ec the Class of Service Concession is modeled as SubClassOf a Public procurement contract, which has its object the right to exploitation the service or both the right to exploitation the service and the payment: <EquivalentClasses> <Class IRI="#ServiceConcession"/> <ObjectIntersectionOf> <Class IRI="#PublicProcurementContract"/> <ObjectSomeValuesFrom> <ObjectProperty IRI="#hasObjectOfContract"/> <ObjectUnionOf> <Class IRI="#RightOfExploitationTheService"/> <ObjectIntersectionOf> <Class IRI="#Payment"/> <Class IRI="#RightOfExploitationTheService"/> </ObjectIntersectionOf> </ObjectUnionOf> </ObjectSomeValuesFrom> </ObjectIntersectionOf> </EquivalentClasses> According to the definition of Art. 3 of the Directive 2004/18/ec the Class of Act of grant to entities others than authority is modeled as SubClassOf a LegalDocument, by which is granted a right (special or exclusive) to a legal person that is not a Contracting authority: <EquivalentClasses> <Class IRI="#ActOfGrantToEntityOtherThanAuthority"/> <ObjectIntersectionOf> <Class IRI="#LegalDocument"/> <ObjectSomeValuesFrom> <ObjectProperty IRI="#byWhichIsGrantedRightTo"/> <ObjectIntersectionOf> <Class IRI="#LegalPerson"/> <ObjectComplementOf>
<Class IRI="#ContractingAuthority"/> </ObjectComplementOf> </ObjectIntersectionOf> </ObjectSomeValuesFrom> </ObjectIntersectionOf> </EquivalentClasses> if then
We have modelled three rules4 using a very preliminary syntax of LegalRuleML5 for testing the emerging standard and providing useful feedback to the OASIS TC: • the rule1 is a defeasible rule and involves this fragment of the Art. 17: “this Directive shall not apply to service concessions as defined in Article 1(4).” The formalization of this text is the following:
X is a member of the OWL class public-procurement-contract X is a member of the OWL class called service-concession X enters in the definition of the Article 1(4) the Directive2004/18/EC shall not apply rule3: “Without prejudice to the application of Article 3” if • the rule2 has strength defeater and is also a meta-rule, since it activates the X is a member of the OWL class public-procurement-contract X is a member of the OWL class called service-concession then the Directive2004/18/EC shall apply the rule3 is activated • the rule3 is related to the content of Article 3; if
X is a member of the OWL class called service-concession
X is a member of the OWL class act-of-grant-to-entity-otherthan-authority then
X shall respect the principle of Non-discriminationClauseOnTheBasisOfNationality
X shall respect of supply contracts which it awards to third parties as part of its activities We concentrate our attention on the connection between rule2 and rule3 using lrml:typeRule=&legalRuleML;metaRule as attribute in the rule2 and the <Ind iri="#rule3">rule3</Ind>: <Rule material="no" id="rule2" lrml:typeRule="&legalRuleML;metaRule"> <!-- Art.17 Without prejudice to the application of Article 3 --> <if id="rule2-body"> <And> <Atom id="rule2-atom1">
<Rel iri="&lkif;#member_of">is a member of the publicprocurement-contract</Rel>
<Var type="&lkif;#public-procurement-contract">X</Var> </Atom> <Atom id="rule2-atom2">
<Rel iri="&lkif;#member_of">is a member of class serviceconcession</Rel>
<Var type="&lkif;#service-concession">X</Var> 4 “Article 17 Service concessions - Without prejudice to the application of Article 3, this
Directive shall not apply to service concessions as defined in Article 1(4).” 5 We take in consideration the version available at the date of the paper submission: http://www.oasisopen.org/apps/org/workgroup/legalruleml/download.php/45888/2.1defeasibility.002.002.doc </Atom> </And> </if> <then id="rule2-head"> <And> <Atom id="rule2-atom2"> <Rel iri="&lkif;#shallApply">shall apply</Rel> <Ind iri="&DIRECTIVE2004_18_ec">Directive2004_18_</Ind> </Atom> <Atom id="rule2-atom2"> <Rel iri="&lkif;#shallApply">shall apply</Rel> <Ind iri="#rule3">rule3</Ind> </Atom> </And> </then> </Rule>
In this paper we have shown how with an hybrid approach that integrates
ontologies in OWL 2.0 (for contents related to legal concepts) and rules in a
preliminary syntax of LegalRuleML (for the dynamic legal knowledge), it is possible
to express in a computable formalism also complex norms, maintaining the
isomorphic [
Up to now, an ontology of European public procurement notices has been developed from scratch in OWL 2.0, in order to represent concepts related to these types of tender documents and processes involved. The ontology is based on both a top-down and a bottom-up approach. Indeed it represents concepts extracted form authoritative sources (Directive 2004/17/EC; Directive 2004/18/EC; etc.) compared with natural language patterns derived from standard forms in use for these tender documents6. The ontology is based on a modular approach and allows, for example, inferences for classifying a contract notice as contract notice with European relevance (i.e. upon EU threshold) or not. Another module of the ontology has the main commitment to classify contract notices covered by the Government Procurement Agreement: through this information it is possible to find out whether a call is open to the participation of economic operators from countries that are not EU members.
RDFa assertions in the XML enable the connection between the structural part of legal texts and the classes of the ontology so developed. Finally upon the ontological level, are represented rules applied to defined ontological classes as well as to materializations inferred through reasoners such as Hermit or Pellet.
We think that future perspectives on this work are strictly related to the investigation of the topic of legal ontology evolution, in order to met the need to automatically detect changes in legal concepts and allow for a sustainable evolutionary system approach. 6 These forms are available at http://simap.europa.eu/buyer/forms-standard/index_en.htm
A particular acknowledgement is due to Enrico Motta, provider of meaningful research inputs during my stage at KMI - The Open University (UK), in the last spring. Thanks go also to the TC of the RuleML Initiative and to the LegalRuleML TC for their helpful hints. I am grateful to my tutor and mentor Monica Palmirani, for her guidance in this challenging research.