Profiles of Legal Knowledge Representation and Reasoning in the Semantic Web: an opportunity for AI in the Public Administration

Abstract

In this paper an approach for legal knowledge representation and reasoning within a Semantic Web framework is presented. It is based on the distinction between provisions and norms and it is able to provide reasoning facilities for advanced legal information retrieval (like implementing Hohfeldian reasoning) and legal compliance checking for deontic notions. It is also shown how this approach can handle norm defeasibility. Such methodology is implemented by decidable fragments of OWL21, while legal reasoning is implemented by available decidable reasoners.

Keywords

• eol>Legal Reasoning
• Semantic Web
• Legal Information Retrieval
• Norm Compliance

1. Introduction

master data (including entities like abstract concepts or real world objects), metadata (namely properties of such Artificial Intelligence (AI) has a very significant impact entities) and their values (reference data). Such RDF not only on human beings’ lives, but most importantly on triples are able to provide a semantic description of a our political, legal and economic institutions. In politics, specific domain: for example, in the legal one, they can AI can support evidence-based rational decision-making, describe facts and legal rules. as well as citizen engagement in policy choices and facil- In the next future the ability of an information sysitate political communication and opinion aggregation. tem to process Linked Open Data (LOD) and to show

In the legal field, a number of models, standards and reasoning capabilities will be essential for developing auapplications are being developed to analyze and classify tomatic legal assistants, endowed with AI capabilities. In documents, apply complex regulations, suggest or predict the legal context the availability of machine readable, acthe outcome of legal cases, detect or anticipate wrongful tionable rules represents therefore a precondition for imconduct, evaluate evidence, analyze sets of legal cases plementing systems with automatic reasoning facilities and social data to detect trends and anticipate changes. for advanced information services. In this contribution In this context, AI methods are inspired by, and combine we present an approach for legal knowledge representawith, the tools and methods of legal theory. The adoption tion and reasoning within a Semantic Web framework. of AI in the legal and socio-political sphere, therefore, It is based on the distinction between provisions and contributes to support the development of efective and norms and it is able to provide reasoning facilities for innovative context-sensitive solutions, thus contributing advanced legal information retrieval (like implementing to democracy and the rule of law. Hohfeldian reasoning) and legal compliance checking for

The semantic web represents one of the main infras- deontic notions. It is also shown how this approach can tructures for AI, as it provides languages for knowledge handle norm defeasibility. Such methodology is implerepresentation and reasoning, as well as smart data for mented by decidable fragments of OWL22, while legal implementing intelligent systems. reasoning is implemented by available decidable reason

RDF1 is the language of the Semantic Web, able to ers. describe a scenario of interest by triples composed by

2. Provisions and Norms

words and sentences, as well as the meaning of such signs. provisions in-force and related norms efective, as well as Following the same twofold view for the legal domain, provisions not in-force and related norms not efective, we can distinguish two levels of interpretation of a lin- we can also have provisions in-force and related norms guistic entity expressing a legal rule: in terms of a set not efective, as well as the symmetric case, provisions of signs organized in words and sentences for creating a not in-force and related norms efective (this last one is normative statement, typically called Provision [ 2 ] [ 3 ], as usually referred as retro-activity (eficacy in the past) or well as in terms of the meaning for application of such ultra-activity (eficacy in the future) of a norm. normative statement, typically called Norm [ 4 ], [ 5 ].

Provision (in-force)

YES YES NO NO

Norm (efective)

YES NO YES3

NO

Having diferent nature, such concepts operate in dif

Provisions have been classified in [ 3 ] in terms of pro- ferent domains. vision types, organised into two main groups (Fig. 1): A provision, as pure textual object, represents the Rules and Rules on Rules. Rules can be Constitutive Rules building block of the legal order (new provisions can as Definition introducing entities, or Regulative Rules as enter or leave the legal order itself). On the other hand, the deontic concepts Duty and Right (or in a more deon- a norm can either modify the text of other provisions (in tic oriented terms Obligation and Permission), as well as case of diferent type of amendments) or can introduce Power etc., regulating subject roles and activities. Rules restrictions on the real world (in case of obligations, for on Rules are diferent kinds of amendments: Temporal, example).

Extension or Content amendments. Each provision type Advanced legal information retrieval, able to impleis characterized by specific properties (for example the ment reasoning on deontic notions, is a type of reasoning Bearer or the Counterpart of a Right), reflecting the law- managing textual information, thus pertaining to provimaker directions. Provision types and properties can be sions. A typical example is a system able to implement considered as a sort of metadata model able to analyti- Hohfeldian reasoning, in which a user submits a query cally describe fragments of legislative texts, hence the to a legal document collection in order to find the rights name of Provision Model [ 3 ]. of a bearer A towards a counterpart B: following an Ho

In this vision, norms represent the way how provisions hfeldian reasoning the system should be able to retrieve are applied; as such they represent the product of an also the provisions expressed as duties of the bearer B interpretative process [ 6 ]. Provisions and related norms towards the counterpart A, because such duty can also have, therefore, diferent roles and properties pertaining be seen as A’s right. An OWL 2 DL approach using the to diferent abstraction levels. Moreover, there may be Provision Model for this type of reasoning is illustrated not a bijective relationship between them: a norm can be in [ 8 ] [ 9 ]. expressed by diferent provisions, as well as it can be valid On the other hand, legal compliance checking is a prothe opposite, namely one provision can include more cess aiming to verify if a fact, occurring in the real world, norms [ 7 ]. They have also diferent relationships with complies with existing norms. Real world scenarios and time. Provisions, as pure textual objects, are the product facts can be efectively represented in terms of ontoloof lawmaking (legal drafting activity and promulgation) gies and related individuals, respectively. Norms, which and are characterized by the in-force date, namely the facts have to be compliant with, provide constraints on starting date of their existence in the legal order. On the reality, therefore they can be modeled as restrictions the other hand, norms are the meaning of provisions, on ontology properties. Such modeling can be used for namely their applicative interpretation; as such they are legal compliance checking. Hereinafter we illustrate an characterized by the eficacy date, namely the starting OWL 2 DL approach for modeling norms and how such date in which a norm can be concretely applied. For modeling can be used for the aim of legal compliance example a taxation rule, in-force at time 1 (time when the checking. related provision enters the legal order), can express the application of a specific tax starting from time 2 (> 1).

In this case 2 is the eficacy date of the norm. Therefore (see Tab. 1), while it is obvious that we can have cases of 3retro-activity / ultra-activity

3. Modeling provisions for advanced legal information retrieval

In [8] and [9] it is shown how Hohfeldian relations on

deontic and potestative notions can be managed within a description logic computational framework. We recall here the main aspects of the approach to show how Provisions can be used to implement an advanced legal provisions retrieval system, endowed with legal reasoning facilities, using a decidable fragment of OWL 2 (in particular OWL 2 DL), therefore exploiting existing decidable reasoners.

In this recall, we show the approach for deontic notions and their relations, sketched in the schema of Fig. 2.4

In order to implement an advanced legal provisions

retrieval system, it is necessary to describe the relations between provisions at the level of the Provision Model. For example the Hohfeldian relation between Duty and Right can be efectively represented by observing that a Right, in correlative correspondence with a Duty, is actually not explicitly expressed in the text, but represents an implicit provision, basically a diferent view of the Duty itself, where the values of the related bearer and counterpart properties are swapped. Therefore, the Provision Model can be extended in terms of Duty and Right5 implicit and explicit disjoint subclasses, able to represent a complete covering of the related superclass (ex: ExplicitRight and ImplicitRight disjoint subclasses represent a complete covering of the Right superclass).

Properties can also be specified as regards both implicit and explicit provisions, so that hasImplicitDutyBearer and hasExplicitDutyBearer are sub-properties of hasDutyBearer, as well as hasImplicitRightBearer and hasExplicitRightBearer are sub-properties of hasRightBearer.

To represent the hohfeldian fundamental relations between Duty and Right, firstly an equivalence relation between their explicit and implicit views is established:

4more details on this modeling approach and its application to potes

tative notions (Power, Liability, Disability Immunity), can be found in [ 8 ] and [ 9 ] 5where “prv:”, namespace for provisions, is hereinafter implied

ImplicitRight ≡ ExplicitDuty and ImplicitDuty ≡ Explic

itRight. In Fig. 3 the established sub-class and equivalence relations between Duty and Right in their explicit and implicit views are summed up.

Moreover, equivalence relations between implicit/ex

plicit Duty and Right properties can be established. In Fig. 4 the asserted properties of ExplicitDuty and ImplicitRight and their mutual equivalence relations are shown (hasImplicitRightBearer ≡ hasExplicitDutyCounterpart and hasImplicitRightCounterpart ≡ hasExplicitDutyBearer).

The same holds for the asserted properties of Im

plicitDuty and ExplicitRight and their mutual equivalence relations (hasImplicitDutyBearer ≡ hasExplicitRightCounterpart and hasImplicitDutyCounterpart ≡ hasExplicitRightBearer) (Fig. 5) .

Note that the proposed patterns do not interfere 4.1. Examples of norms representation with the relations between Right and Duty, which and compliance checking still hold. In fact, for the couple Right/Duty, an individual of ExplicitDuty is also an individual of Duty, Let’s consider as example the rule R1. The related scegiven the axiom rdfs:subClassOf(ExplicitDuty, Duty). nario can be modeled in terms of an ontology including Moreover the axiom owl:equivalentClass(ImplicitRight, a class Supplier, having a boolean property hasComExplicitDuty) tells us that such individual is also an municatedConditions. Norm R1, expressing a duty for ImplicitRight, which is also a Right, given the axiom the suppliers states that suppliers must communicate rdfs:subClassOf(ImplicitRight, Right). Since this is done contractual terms and conditions to the consumers: the symmetrically for explicit and implicit duties and rights, individuals of the class Supplier complying with this we can deduce that Right is equivalent to Duty, namely norm are all those ones belonging to the subclass Suppliis another reading of the Duty itself, given that the union erR1Compliant identified by a restriction on the boolean of the disjoint explicit and implicit subclasses covers com- property hasCommunicatedConditions to have value pletely the related superclass. “true” (see Fig. 6, where myo: is a fictitious namespace representing MyOntology). 3.1. Example of provision representation

and reasoning

In order to show the ability of the Provision Model ap

proach to provide advanced legal information retrieval facilities, based on provisions and related hohfeldian relations, the following example of a legal rule R1 can be used: R1 : The supplier shall communicate to the consumer all the contractual terms and conditions

In terms of the Provision Model, this rule can be seen as a provision of type Duty, which can be represented as

ExplicitDuty(Supplier, Consumer), where the arguments of the ExplicitDuty are the explicit bearer (Supplier) and related explicit counterpart (Consumer), respectively. Given the following introduced hohfeldian relations:

ImplicitRight ≡ ExplicitDuty ImplicitRightCounterpart ≡ ExplicitDutyBearer ImplicitRightBearer ≡ ExplicitDutyCounterpart

the provision at R1 can also be seen as ImplicitRight(Consumer, Supplier), including related implicit right bearer (Consumer) and implicit right counterpart (Supplier). Therefore, the provision R1 can be retrieved asking for either the duty of the supplier or the right of the consumer.

4. Modeling norms for legal compliance checking

As discussed in Section 2, norms can be viewed as the application of legal provisions, providing constraints on a real world scenario to be regulated. In the Semantic Web a real world scenario is usually represented by a domain ontology. In this context a norm, providing constraints to such scenario, can be modeled in terms of constraints on the domain ontology: for example, in case of obligations (like a duty), as ontology property restrictions.

Such a representation for the real world scenario and

related norm expressed by R1 results in the OWL 2 DL, decidable profiles. This allows us to use a OWL 2 DL decidable reasoner in order to implement reasoning facilities, preparing the ground for compliance checking with respect to R1. The inferred model establishes a rdfs:subClassOf relationship between SupplierR1Compliant and Supplier (as shown in Fig. 6), where SupplierR1Compliant is the class of all the individuals of type Supplier having “true” as value of the property hasCommunicatedConditions. Therefore, compliance checking according to the norm R1 is a problem of checking if an individual of type Supplier belongs to the class SupplierR1Compliant.

As an example let’s consider the following two individuals myo:s1 and myo:s2 of the class Supplier: myo:s1 is an individual not compliant with R1, while myo:s2 is complaint with R1. The following SPARQL query

SELECT ? x WHERE { ? x r d f : t y p e myo :

S u p p l i e r R 1 C o m p l i a n t } is able to select the individuals which are complaint with R1 (in our case s2). Legal reasoning in terms of norm compliance checking is therefore performed within a decidable computational complexity profile. 4.2. Norm compliance and defeasibility

In this section we show how the presented approach for norm representation and compliance checking can handle norm defeasibility. Let’s consider the following legal rule R2:

R2 According to a [country] law one cannot drive over 90 km/h

In the case of R2, the vehicles circulation scenario

can be modeled in terms of an ontology including a class Driver, having a datatype property hasDrivingSpeed with range in the xsd:float datatype.

Norm R2, expressing an obligation on the vehicles circulation scenario, states that, according to the related country law, one cannot drive over 90 km/h: the individuals of the class Driver complying with this norm are those ones belonging to the subclass DriverR2Compliant having value ∈ [0.0, 90.0] Km/h on the datatype property hasDrivingSpeed (Fig. 7).

In other terms the norm R2 is represented as restriction on the property hasDrivingSpeed able to identify the class DriverR2Compliant which is equivalent to the class of the individuals for which the values of the property under consideration are in the range [0.0, 90.0] km/h. In R2 : According to a [country] law, one cannot drive over order to represent such constraints the following restric- 80 km/h tion on the datatype property myo:hasDrivingSpeed to values (inclusively) between 0.0 and 90.0 can be expressed The new version of R2 (Fig. 9) can defeat the previous by the xsd:minInclusive and xsd:maxInclusive datatype compliance conclusions, in the sense that individuals, bound properties. Such a representation results in the which were compliant with the old version of R2 (Fig. 8), OWL 2 DL decidable profile. might not be compliant with it anymore (this is the case

As in the previous example, the inferred model in the example in Fig. 9 of the individual d3). In order to establishes a rdfs:subClassOf relationship between cope with this change, the same model can be updated DriverR2Compliant and Driver (as shown in Fig. 7), (without changing anything on the names of the classes) where DriverR2Compliant is the class of all the individ- just by changing the original restriction on the datatype uals of type Driver having values of the property has- property hasDrivingSpeed with a new one expressed DrivingSpeed in the interval [0.0, 90.0] km/h. Therefore, by the new version of R2, as shown in Fig. 9. Without compliance checking according to the norm R2 is a prob- changing anything on the individuals, their membership lem of checking if an individual of type Driver belongs to the class DriverR2Compliant changes accordingly so to the class DriverR2Compliant. that, for example, the individual d3, compliant with the

As a concrete example, let’s consider four individuals old version of R2 (Fig. 8), is no more compliant with the myo:d1... myo:d4 of the class Driver, as represented in new version of R2 (Fig. 9). Therefore, the query able to Fig. 8. select compliant individuals remains the same:

In this list of individuals, the individual myo:d4 is not

compliant with R2 (having speed 95.0 Km/h ≥ 90.0 Km/h). The following query: is able to select the individuals which are complaint with R2 (in our case myo:v1, myo:v2, myo:v3).

Using the same example, we can now show how this compliance checking modeling approach can cope with norm defeasibility. Defeasibility is the property of an argumentation system for which a conclusion is open to revision in case evidence to the contrary is provided [ 10 ]. This particularly holds in legal reasoning which is a typical case of non-monotonic reasoning, where norm conflicts or norm exceptions might breach a previous conclusion.

Let’s consider rule R2, as previously modeled, and the following new version of rule R2, introducing a more strict driving speed limit at 80 Km/h: which is able to retrieve the only individuals d1 and d2, compliant with the new version of R2.

5. Conclusions and future developments

In this paper we have presented a legal reasoning ap

proach based on the distinction between the concepts of provisions and norms, able to deal with diferent types of legal reasoning, in particular advanced legal information retrieval, as well as norms compliance checking. The method is based on the use of decidable fragments of OWL 2, able to guarantee the computational tractability of the approach. This represents an essential property of a legal reasoning system in the Semantic Web, characterized by a huge amount of Linked Open Data in the form of triples.

Nowadays, within the public administration, AI is no longer just theory, but it’s becoming an increasingly important option. Moreover, the Public Administration (PA) may play a central role in AI systems development, because they produce a large amount of public data, and legal data in particular, whose accessibility and reuse can be improved by applying semantic web technologies, as shown in this paper. This aspect is considered so strategic for business and public administration that each European country has developed its own national strategy. Applications for public administrations are aimed to create data infrastructures able to exploit the potential of big data that the PA generates, to simplify and personalize the ofer of public services and to innovate administrations.

References

[1] J. Searle , Speech Acts: An Essay in the Philosophy of Language , ISBN 978-0521096263 , Cambridge University Press, 1969 .

[2] J. Raz , The Concept of a Legal System , Oxford University Press, 1980 .

[3] C. Biagioli , Modelli Funzionali delle Leggi. Verso testi legislativi autoesplicativi ., volume 6 of Legal Information and Communications Technologies Series, European Press Academic Publishing, Florence, Italy, 2009 .

[4] R. Guastini , Le Fonti del Diritto . Fondamenti teorici ., Giufrè , Milano, 2010 .

[5] A. Marmor , The Language of Law, 978-0-19-871453- 8 , Oxford University Press, 2014 .

[6] H. Kelsen , General Theory of Norms , Clarendon Press, Oxford, 1991 .

[7] G. Pino , Teoria analitica del diritto , 9788846744517 , ETS, 2016 , pp. 144 - 183 .

[8] E. Francesconi , Semantic model for legal resources: Annotation and reasoning over normative provisions, Semantic Web journal: Special Issue on Semantic Web for the legal domain 7 ( 2016 ) 255 - 265 .

[9] E. Francesconi , A description logic framework for advanced accessing and reasoning over normative provisions , International Journal on Artificial Intelligence and Law 22 ( 2014 ) 291 - 311 .

[10] T. Athan , G. Governatori, M. Palmirani , A. Paschke , A . Wyner, LegalRuleML: Design principles and foundations , in: The 11th Reasoning Web Summer School , DOI: 10.1007/978-3- 319 -21768- 0 _ 6 , 2015 .