Jusletter IT

Compliance control is a demanding task for public administration, accounting for most manual case handling activities inside public administration, and a significant amount of work for citizens and businesses.

Tax administrations, immigration services, and social security administrations cannot possibly check every individual transaction for compliance. Information comes at a price. People may be in a position to know and a position to inform the administration, but the costs of producing information are not always proportional to the information value. Wage data is for instance periodically reported by employers, and used as a basis for wage tax, social security premiums, benefits, etc. Checking actual contracts, payments, and work, requires intensive observation, inspection of written evidence, the expensive presence of the employers’ accountant, and causes significant inconvenience and costs. Inspections based on concrete suspicions are equally intrusive and unfairly distributed, even if the suspicions are well-founded, because innocent third parties regularly incur costs for allowing the tax administration to check the conformance of others.

Governments regularly express a need for better service alignment and data reuse, and try to reduce administrative burdens, for instance by limiting regular data reporting obligations. In public administration many are convinced this comes with a price: less data for compliance control. The reality is, however, that we know little of the quality of compliance controls, because compliance is addressed fragmentarily, based on domain-specific economic microtheories, or complaints, public scandals, and case law [1,2]. Ideas about types, indicators, and prevalence of noncompliance scenarios in public administration are often not explicit, not effectively shared between civil servants, and not tested with proper scientific controls. Data reporting intervals, like the fiscal year, moreover limit visibility and enforceability of nonconforming behaviours.

Public administration is in need of a systematic approach to integrating these various storylines and microtheories into a coherent design, monitoring, and diagnosis problem space for compliance controls. Story-based knowledge representation approaches for case law are attractive for knowledge acquisition from case law and experts, but do not readily translate into a design search space for compliance controls. This paper proposes a connection between these stories and a design space for compliance controls in public administration. Section 2 of this paper sketches some of the context of this work. Section 3 presents detailed motivating examples of the compliance controls problem. In section 4, a simple knowledge acquisition framework for noncompliance stories is presented, and the compliance controls problem is split into various, competing parametric design frames.

To comply is to intentionally, systematically, conform to the rules. Compliant organizations are expected to put in an effort to ensure that personnel and clients are aware of, and take steps to, comply with relevant laws and regulations. Noncompliance is the opposite of compliance: intentional, systematic evasion of a rule. The intentions ascribed to actors in a story matters for the reaction. It is a tacit assumption that one should know what rules to comply with, but unfamiliarity with applicable rules and differences in interpretation can lead to unintended noncompliance. Groups of people may show systematic noncompliance, while participating individuals do not understand the big picture.

Public administration is actively and passively monitoring for noncompliance, and replying, to the individual case and to the observed noncompliance pattern. Classifying a case as noncompliant is the result of a (legal) qualification process for which one needs evidence and, in legal or administrative processes, formal power to decide on the case.

In [3] we described how modeling noncompliance stories and translating them into generalized classification patterns is part of a design process, while the activation and deactivation of classification patterns, and information gathering and classification based on those patterns, are part of operational processes. In addition, [3] addressed the use of generalized agent role models to explore microtheories in multi-agent simulation. We also noted in [3,4] that the categorization of the resulting AgentSpeak scripts, representing normal and abnormal behaviours in a role, was still an open problem. In this paper the problem is addressed, by proposing a simple parametric problem solving framework based on the cycle in [3], that already suggested a categorization of scripts by their role in problem solving.

In the field of computer science & law, there are two accounts of stories that may be considered direct precursors to the conceptualization presented here. There is firstly, a tradition of story-based inference to the best explanation, where the object is to match the evidence presented in a case with the best story, constructed from causal generalizations, to explain it (e.g. [5,6]). Bex et al. in [7] presents a hybrid proposal with both causal and evidentiary inference. The reusable design components in this conceptualization are rather small: defeasible evidentiary and causal generalizations that may be used to back up individual inferences. While this is a flexible theoretical model of the construction of hypothetical stories, it presents practical problems for knowledge management in a large organization that knows of many stories. Secondly, Hoekstra et al. in i.a. [8] propose the allocation of hypothetical mental processes to agents as an account of common sense causal explanation of stories. Attractive in this proposal is that it applies the model-based reasoning paradigm to mental models.

This paper proposes a classification approach to this mental process allocation problem, allocating complete behaviour scripts to the agents in a story, to find a best explanation. This approach reflects the intuition that mental processes, as causes of events, generally come in coherent packages constrained by social role expectations, taught behaviours, and the general requirements of rationality.

Suppose that we want to obtain enough input information to make a certain decision. The information needed may be obtained by performing a sequence of actions that yield information. Some information has already been obtained, suggesting certain hypotheses. Suppose that we, recursively, perform one additional action and evaluate the resulting information and the earlier information to discard hypotheses, until we are ready to make the decision.

The compliance/conformance decision making context is a diagnosis problem, and the associated information collection problem is a monitoring problem [4]. From the point of view of the compliance monitor, the objects of analysis are noncompliance stories. Abstract storylines are associated with a legal qualification backed by a legal rule: for instance client nonconformance, taxable event, internal performance problem, undecidable case, etc. A storyline consists of (a sequence of) events that happened, involving a number of agents that sent messages to each other, caused by, and resulting in, mental processes. Agent role behaviour scripts add the plan operators that causally connect the events, supplying both a rationale for the events that happened, and a basis for allocation of responsibility for the qualified events to specific agents. This responsibility assignment is a diagnosis in a narrow sense. An observation is essentially a received message by the diagnostic agent, caused by an action of the diagnostic agent. A sequence of observations may be consistent with a storyline, but should be logically distinguished from the events that make up the story, in which the diagnostic agent (usually) plays no role.

Essentially, what we are seeking is a agent role behaviour script assignment that explains the observations given a background theory. That is, the agent role assignment is consistent with the background theory and observations, and the observations follow from the background theory and the agent role assignment. This problem definition follows the familiar inference to the best explanation problem definition [5], but in a very restricted form: the set of abducibles only consists of assignments of behaviour scripts to the agent roles. As we argued in i.a. [4], this static problem definition can be interpreted as model-based diagnosis as formalized by e.g. Reiter in [9].

A diagnosis is a hypothesis that a subset of the agent roles is abnormal and the rest normal. As we argued in [4], default reasoning about agent behaviour can be modelled with a normal default theory [9]. For this default theory, Reiter’s default logic extensions are those of the generic diagnosis problem directed towards a minimal set of abnormal components [9]. Reiter essentially treats model-based diagnosis as a simple parametric design framework, where the main object is to allocate the qualifications {healthy, faulty} to components in a structure, meeting the minimality requirement.

To recommend this as a generic diagnosis policy would suggest that diagnostic information value from a default reasoning perspective is a leading principle in implementing compliance controls. This is not usually the case. Monitoring policies are justified with various pragmatic, statistical, and qualitative arguments, and not inferred from diagnostic information value, i.e. their effect on navigation of the search space, on a case-by-case basis.

The logic guiding the selection of actions that may expand the set of observations is complex and problematic. A number of factors plays a role in selection of the action:

1. the diagnostic value of information,
2. the reliability of the information that can be obtained,
3. the costs of the action to obtain it, and
4. the effects of obtaining it on future action.

If we consider this problem with mainly the interplay between cost, value, and reliability in mind, and assume that these do not, or very rarely, change in time, the problem in principle lends itself well to automated learning techniques that can be implemented in an organization, based on information value theory (cf. generally [10]). Cost, value, and reliability do however change in time, and the long term effects of compliance monitoring policies are hard to predict reliably.

A first problem that is especially acute in social domains, is that the use of predictable monitoring policies teaches the people that are being monitored to evade compliance controls, actively undermining assumptions about the value, cost, and reliability of information. Policy affects the domain, and not always in directions intended by the policy maker.

A second, general problem with policies to obtain information is reject inferencing. Public administration rarely undertakes large-scale monitoring experiments for explorative purposes. The courts may moreover reject monitoring policies that impose significants costs on clients (like taxpayers) for mere explorative purposes, with reference to equality, subsidiarity, or privacy. Statistical optimization of monitoring policies in the absence of statistical controls leads to impoverished knowledge of the behaviour of populations that are never targeted for monitoring, invalidating the statistical basis of the monitoring policies.

Finally, the minimality assumption is questionable: in an abnormal transaction, it may well be usually the case that both behaviour scripts involved are abnormal.

In the next section, we illustrate the factors that influence the quality of monitoring policies with two examples.

The train conductor stands in the doorway of his train, watching a fight between two men erupt on the railway platform. While he routinely picks up the microphone to warn the passengers that pickpockets may be present in the train, a passenger walks up to him: «My laptop bag has just been stolen while I was watching those men from the window».

This is a familiar storyline: the pickpocket’s diversion. The story is consistent with a number of very different hypotheses, and police officers will have a hard time producing evidence:

1. The pickpocket used a spontaneously arising diversion to steal the bag, making the pickpocket the sole offender.
2. An accomplice of the pickpocket picked a fight on the platform to create a diversion, allowing the pickpocket to steal the bag. The pickpocket and one of the two men involved in the fight are co-offenders.
3. Two accomplices of the pickpocket simulated a fight on the platform to create a diversion, allowing the pickpocket to steal the bag. The pickpocket and both men involved in the fight are co-offenders.
4. The passenger, guessing how the fight will be interpreted by the train conductor, uses the opportunity to make the train conductor a witness to the simulated theft of a non-existent laptop, in order to make an insurance claim. If the train conductor calls the railway police, the insurance claim of the passenger gains credibility. The passenger is the sole offender.
5. The passenger left it on the railway platform. No offense took place. Perhaps the laptop bag was brought to a lost and found desk.

Immediate collection of witness statements, in the train and on the railway platform, would help the police to pick the most likely scenario. But the train must go on, and a minute later most witnesses will have left the scene of the crime. Suppose that you are a railway police officer hurrying to the scene before the train leaves. What do you do? Do you want to board the train or stay on the platform? Are the two men on the platform, the thief, or the passenger in the train on your mind?

What should a railway police protocol for this type of event look like? And, considering the paucity of evidence: Does this kind of scenario ask for cameras in trains and on platforms? Should the railway company store personalized ticket information? This is the kind of question that public administration faces in implementing compliance controls. The legal qualification of the story – theft, assault, insurance fraud – and the agents in the story that should be held legally responsible, depend on which scenario really happened. Both the qualification and the assignment of responsibility largely depend on the intentions attributed to agents, even if those intentions do not later play a formal evidential role. Note that the thief is not physically present in the scene, just presumed to exist. Note that timing is important for collecting evidence.

Now consider this more relevant storyline:

The cadastral register claims (e₁) that a project developer sold an apartment for €350,000 to a natural person (e₂). The payroll tax register claims (e₃) that the project developer paid wages to that person (e₄). The cadastral register claims (e₅) that this person sold the apartment three months later for €500,000 to a third person (e₆), but NVM real estate brokers claims (e₇) that median home price hardly changed in that period (e₈).

This scenario is relevant to a tax administration. The story is consistent with a number of very different hypotheses, which are often hard to tell apart. In the field of real estate we find a variety of types of crime consistent with the observation of large deviations from a reasonable market price, or untypical quick depreciation or appreciation of a property. People may sell below or buy beyond a reasonable market price to avoid income taxes; in a foreclosure auction, a seller may unknowingly or unwillingly sell his property below a reasonable price to a buyer cartel, which distributes the profit among the cartel participants (bid rigging); the transaction may hide a bribe or theft; and finally the seller or buyer may be a victim of extortion.

The reader of this paper, having his imagination already activated by the previous example, will undoubtedly realize that the information presented permits for many detailed hypotheses of varying plausibility, for instance:

1. NVM median home price development is not representative for the specific property.
2. The project developer, in order to avoid immediate insolvency, sold a property in great haste to one of its employees.
3. The employee, acting as the agent of his employer, stole €150,000.
4. The employer paid the employee €150,000, avoiding payroll taxes.
5. The third person was forced at gunpoint to acquire a property worth €350,000 for €500,000.
6. The employee, acting as the agent of his employer, was forced at gunpoint to cash in €150,000 from his employer, and an extortionist made of with the money.
7. Another employee, who owed the first €150,000 euro, acting as agent of the employer, sold his property to the employer first, in order to extend the transaction chain and escape detection, and then to the other employee, who cashed in the difference, to evade income taxes.
8. The project developer was a victim of bid rigging in a foreclosure auction, coordinated by the employee.

 

We can’t decide, based on the story alone, whether something extraordinary happened, and, if so, who is the beneficiary of the extraordinary transaction. Different explanations of the story point to different offenders and victims, and to different legal qualifications of the story. Note also that the value of witness testimony depends on these qualifications: an offender for instance has a prima facie motive to misrepresent what happened.

The events in the story are the result of the plans of the agents involved. Rather than understanding the story directly on the level of plans, we propose to explain the story on the level of agent role behaviour descriptions for the buyer and seller. This allows us to group together in a coherent unit the goals and adaptive plan operators that characterize

1. plans for buying or selling, and
2. plans for dealing with information seeking by third parties.

This deals with the intuition that the answers to the classical critical questions about the reliability of testimony depend on the roles one could play in the story. These critical questions are whether the agent is in a position to know something, and whether the agent has a motive to falsely claim something. In addition, we have a way of dealing with track covering behaviours triggered by information seeking.

Let’s consider the example story in this context. If it is a simple case of tax evasion, the buyer and seller collude against the tax administration, and will coordinate testimony. They have reason to fabricate evidence for circumstances that would make the transaction normal, for instance by having the property retroactively appraised for €350,000 by a fraudulent professional appraiser, claiming great urgency, or claiming that the sale was agreed long ago when the reasonable market value of the property was €350,000. In the case of a theft, on the other hand, an agent representing the seller and the buyer colluded, but another agent representing the seller would likely offer reliable testimony. It is arguable that the agents were in a position to know that the suspicious transaction was going to be easily noticeable, and that therefore something more complicated, involving for instance extortion, could be going on. A thief or extortionist, becoming aware of having aroused suspicion, may react by threatening potential witnesses. Etcetera. These processes all influence the effectiveness of the monitoring process.

The following two behaviour descriptions for tax evasion satisfactorily explain event e₂. e₄ is only potentially relevant as a background to the motive for paying the employee. Plan operators follow the general pattern event (given conditions) ← plan, and adopted goals (intentions) in events and conditions start with to. First we present a script for the seller s:

1. Goal: to pay an untaxed amount a to b.
2. To pay an untaxed amount a to b ← propose to sell a property worth some v for v – a to b pretending the agreement was made long ago.
3. b accepts proposal to sell a property worth some v for v – a←propose a specific property worth v to b.
4. b accepts proposal of a specific property being worth v ← secure property for v then sell property for v – a to b.
5. To sell property for v – a to b← offer to sell property for v – a to b.
6. b accepts offer to sell property for v – a ← register property transfer to b in cadastral registration then monitor payment of v − a by b.
7. Someone asks about the price v – a given the goal to pretend the agreement was made long ago ← claim the agreement was made verbally long ago.

In the interest of simplicity the plan is rather static. Note (2, 7) that fabrication of evidence plays a role in the planning. Next we present a script for the buyer b:

1. Goal: to receive an untaxed amount a from s.
2. s proposes to sell a property worth some v for v − a pretending the agreement was made long ago given the goal to receive an untaxed amount a from s ← accept proposal from s to sell a property worth some v for v − a.
3. s proposes a specific property worth v to b ← have property for v appraised then consider proposal of specific property for v.
4. Property for v was appraised given the goal to consider proposal of specific property for v and given that the appraisal shows that the property for v is acceptable ← accept proposal of the specific property being worth v.
5. s offers to sell property for v − a to b given accepted proposal of a specific property being worth v ← accept offer to sell property for v − a then monitor registration of property transfer by s in cadastral registration then pay v − a to s.
6. Someone asks about the price v − a given the goal to pretend the agreement was made long ago← claim the agreement was made verbally long ago.
7. Someone asks about the appraisal given the goal to pretend the agreement was made long ago ← claim the agreement was an option to purchase, and the appraisal was made to decide about exercising the option then quickly inform s about the optional nature of the agreement.

Here again fabrication of evidence plays a role (2, 6, 7). In this plan, having the appraisal performed may attract attention: why have the property appraised before acceptance if the agreement already exists? Step 7 is a plausible, but – from a timing point of view – hairy improvisation in response to the monitor, that must be supported by s.

The associated monitoring script interacts with the cadastral registry, payroll data registry, online NVM database, appraisers, buyers, and sellers. It is omitted for reasons of space, considering its complexity even in a simplified domain model. Let us assume the investigation completes the story with the following interpretation of events by the monitor:

The parties involved agree (e₉, e₁₀, e₁₁) that the reasonable market price is €500,000 (e₁₂). The project developer (e₁₃) and its employee (e₁₄) both claim to have agreed previously to an option to purchase the property for a then reasonable price (e₁₅). The employee later also claimed (e₁₆) that the agreement was an option to purchase, and the appraisal was made to decide about exercising the option (e₁₇). The tax administration claims (e₁₈) that the verbal agreement claimed in e₁₅ is an unverifiable result of a fabrication for the purpose of evading taxes (e₁₉). The project developer and employee countered with the claim (e₂₀, e₂₁) that the relationship of trust that exists between them made requiring a written agreement imprudent (e₂₂). The tax administration claims (e₂₃) that the difference between the reasonable price and the actual price paid (€150,000) is taxable income (e₂₄).

From an explanatory point of view, e₂₄ is the conclusion, supported by e₁₂ (e₉, e₁₀, e₁₁) and e₂ (e₁). The argument for a contrary conclusion is supported by e₁₅ (e₁₃, e₁₄). If e₁₉, the weak attack in the tax administration’s reconstruction on e₁₅, is not accepted by the tax court, a new policy guideline points the way out: only written options to purchase real estate should be enforceable, and they should be duly registered somewhere to prevent antedating. More generally, the tax administration’s explanation of events is that the project developer and employee were following scripts s and b, and that the tax administration’s specific monitoring policy in this case, script m, and the opportunities created by the civil code’s rules on agreements, which require that real estate transactions must be provable to third parties, but options to purchase need not be, makes enforcement of tax evasion a problem in this case. A note of caution: this plan is obviously not state-of-the-art in tax evasion.

In [3] we discussed the use of AgentSpeak scripts representing typical behaviours, and noted that finding a logical organization of such scripts is an open problem. In this presentation, which is about inference, we structure scripts in such a way that it is easy to present specific parametric design and monitoring problems in an ontology as an assignment of instances to classes. The generic components of the knowledge acquisition framework are the following:

1. A set of agent role instances ARI = {ari₁, ari₂, . . . , arin}
2. A set of agent role classes ARC = {arc₁, arc₂, . . . , arcn}
3. A complete 1:n relation arc(ari) of agent role classes to instances (for all ari ∈ ARI there is an (ari, arc) ∈ ARIC where arc ∈ ARC)
4. A set of normal behaviour scripts for each agent role class, NS(arc) = {ns₁, ns₂, . . . , nsn}
5. A set of abnormal behaviour scripts for each agent role class, ANS(arc) = {ans_1<, ans₂, . . .., ansn}
6. A sequence of events E = {e₁, e₂, . . . , en}, where events are messages m(a₁, a₂, p), as given before in section 2, and p may again be a proposition about a message.
7. Some arbitrarily structured background theory BT that permits consistency checking.

 

An agent role behaviour script allocation is an n:1 relation ALLOC, where for all (ari, ns) ∈ ALLOC it is the case that ari ∈ ARI, arc(ari) ∈ ARIC, and either ns ∈ NS(arc) or ns ∈ ANS(arc). A complete allocation allocates a script to all ari ∈ ARI. An allocation is an explanation of E, given BT, if E follows from ALLOC∪BT, ALLOC does not already follow from BT ∪ E, and ALLOC ∪ BT ∪ E is consistent (cf. for instance [5] for compatible accounts). Explanations need not be complete, and a partial allocation may form part of the problem definition.

The difference between a an explanation, a diagnosis, and a parametric design is what one does with the output.

A diagnostic explanation of E is an allocation ALLOC, where, for at least one (ari, ns), ns ∈ ANS(arc). This does not necessarily mean that all non-diagnostic explanations do not fit. A diagnosis suggests a repair: a diagnostic explanation leads to a legally or administratively relevant and actionable qualification of the behaviour. This is a problem domain in itself, not addressed here. That there is only a diagnosis problem if all non-diagnostic explanations have been rejected is a simplifying assumption in problem solving frameworks, not a reflection of reality. The ultimate objective is not to reach an unambiguous conclusion, but simply to select a next action.

A parametric design for the agent roles inside the organization meets all requirements of an explanation (cf. for instance [11] for parametric design problem solving). Sequences of events no longer function as a thing to be explained, but as design requirements. Performance criteria often are part of the problem formulation. In line with our general approach, performance qualifications should be causing observable events.

Actual design and explanation problems, as dealt with in public administration, depend on splitting the set of agent role instances into several subsets. There is the monitoring system (MON ⊂ ARI) that is monitoring, and the domain (DOM ⊂ ARI\MON) being subjected to monitoring. In the domain DOM in public administration we find clients and network partners outside (OUT ⊂ DOM, NET ⊂ OUT, CLIENTS ⊂ OUT), and we find case handlers on the work floor inside the organization (IN ⊂ DOM). Compliance monitoring seeks an explanation for CLIENTS, or OUT, or DOM, while other script allocations are part of the background theory.

In the design of compliance monitoring subsystems we can work with specific script allocations for CLIENTS, or OUT, orDOM, for instance suggested by case law, past case data, expert knowledge, or informed guesses based on for instance statistics. In this case the desired conclusions are known, and an allocation of monitor behaviour scripts is sought that reproduces it. The design of compliance controls involves the entire IN set, and, supported by policy making or service agreements, will also address so-called network partner agents in the NET set, like the cadastral registry, NVM, and appraisers in section 3.

Legal practitioners depend on a lot of common ground, expressed in jurisprudence and case law, for the justification of individual legal decisions. While it is clear that judgment of a policy cannot be a mere derivative of judgments on individual decisions that the policy produces, jurisprudential frameworks only address the legal justification of case handling and monitoring policies in public administration in generic terms.

This paper describes a way of dealing systematically with noncompliance storylines and an associated compliance controls design space. The agent role-centered approach, even though the parametric design space proposed here has obvious limitations, effectively organizes knowledge about causal mechanisms mediated by mental processes in noncompliance stories, and helps designers to gain insight in the compliance controls design space. The approach is an alternative to problematic domain ontology approaches in public administration knowledge management. It is complementary to domain-specific and labour-intensive economic simulation models in fields like fiscal law, and story explanation approaches in law like [7,8].

The work presented in this paper is an extension to our work in [3,4]. The story explanation and compliance controls design space specification is based on an analogy with parametric design problems, which are well-understood in computer science.

[1] Verschueren, J.: The impact of legislation: a critical analysis of ex ante evaluation. Martinus Nijhoff publishers (2009)

 

[2] Gribnau, J.L.M., Lubbers, A.O., Vording, H.: Terugkoppeling in het Belastingrecht. SDU (2008)

 

[3] Boer, A., van Engers, T.: An agent-based legal knowledge acquisition methodology for agile public administration. In Ashley, K.D., van Engers, T.M., eds.: The 13th International Conference on Artificial Intelligence and Law, Proceedings of the Conference, June 6–10, 2011, Pittsburgh, PA, USA, ACM (2011)

 

[4] Boer, A., van Engers, T.: Diagnosis of multi-agent systems and its application to public administration. In Abramowicz, W., ed.: Business Information Systems Workshops – 14^th International Conference, BIS 2011, Poznan, Poland, June 15–17, 2011. Volume 97 of Lecture Notes in Business Information Processing, to appear., Springer (2011)

 

[5] Lipton, P.: Inference to the best explanation. Philosophical issues in science. Routledge (1991)

 

[6] Keppens, J., Schafer, B.: Knowledge based crime scenario modelling. Expert Syst. Appl. 30(2) (2006) 203–222

 

[7] Bex, F.J., van Koppen, P.J., Prakken, H., Verheij, B.: A hybrid formal theory of arguments, stories and criminal evidence. Artif. Intell. Law 18(2) (2010) 123–152

 

[8] Hoekstra, R., Breuker, J.: Commonsense causal explanation in a legal domain. Artif. Intell. Law 15 (January 2007) 281–299

 

[9] Reiter, R.: A theory of diagnosis from first principles. Artificial Intelligence 32(1) (1987) 57–95

 

[10] Howard, R.: Information Value Theory. IEEE Transactions on Systems Science and Cybernetics 2(1) (1966) 22–26

 

[11] Motta, E., Zdrahal, Z.: Parametric design problem solving. In: Proceedings of the 10th Banff Knowledge AcquisitionWorkshop. Number 1990, SRDG Publications, University of Calgary (1996) 1–24

 

 

Alexander Boer. Leibniz Center for Law, University of Amsterdam, The Netherlands. E-mail: A.W.F.Boer@uva.nl.

 

Tom Van Engers. Leibniz Center for Law, University of Amsterdam, The Netherlands

 

This article is republished with permission of IOS Press, the authors, and JURIX, Legal Knowledge and Information Systems from: Kathie M. Atkinson (ed.), Legal Knowledge Systems and Information Systems, JURIX 2011: The Twenty-Fourth Annual Conference, IOS Press, Amsterdam et al.