Formalising Jurisprudence / Formalisierung der Jurisprudenz E-Book

Friedrich Lachmayer

• Herausgeber: Editions Weblaw
• Kategorie: Fachliteratur
• Sprache: Englisch

Seitenzahl: 566

Veröffentlichungsjahr: 2019

Beschreibung

Hajime Yoshino ist seit den Siebzigerjahren ein maßgeblicher Protagonist der Rechtslogik und der Rechtsinformatik in Japan. Der Gelehrtencommunity um Herbert Fiedler, Arthur Kaufmann, Ulrich Klug, Lothar Philipps, Jürgen Rödig, Ilmar Tammelo und Ota Weinberger entstammend, hat er sich unermüdlich für die Anwendung der Logik in der Rechtswissenschaft eingesetzt und damit auch den Weg für die Rechtsinformatik mitbereitet. Sein Anliegen, das er in der "Logischen Jurisprudenz" zusammengefasst hat, ist Auftrag und Zeichen, die formalen Wurzeln der Rechtswissenschaft im Übergangsfeld zur Rechtsinformatik weiterhin zu verstärken und auszubauen. Professor Hajime Yoshino ist eine angenehme und humorvolle Persönlichkeit. Es ist ein Privilegium, ihn treffen zu können. Er regt die Diskussionen an und wirkt integrativ. Im Zuge der Gestaltung dieses Sammelbandes wurde erneut deutlich, dass es in der wissenschaftlichen Praxis nicht nur um den Kernbereich der Anwendung der formalen Logik geht, sondern dass das Wort "Formalisierung" ein weiteres Feld beschreibt, das – der Avantgarde zugeordnet – durch explizite Strukturierungen eine intellektuelle Durchdringung des Rechtes und seines Umfeldes aufbereitet. Die Vielzahl der in diesem Sammelband behandelten Themen gruppiert sich in unterschiedlicher Intensität um Yoshinos Anliegen einer Rechtswissenschaft, in welcher der Logik und dem formalen Denken ein grundlegender Stellenwert zukommt.

Leseprobe

Formalising Jurisprudence

Festschrift for Hajime Yoshino

Formalisierung der Jurisprudenz

Festschrift für Hajime Yoshino

Erich Schweighofer / Michał Araszkiewicz /

Friedrich Lachmayer / Marijan Pavčnik (Eds. / Hrsg.)

Formalising Jurisprudence

Festschrift for Hajime Yoshino

Formalisierung der Jurisprudenz

Festschrift für Hajime Yoshino

Liber amicorum

Editions Weblaw, Bern 2019

OCodex

ICommentatio

IIColloquium

IIIDissertatio

IVDoctrina

V Liber amicorum

VIMagister

VIIMonographia

VIIIThesis

IXScriptum

XAnthologia

Editions Weblaw

Bestellung und Vertrieb Schweiz: Weblaw AG, Bern

Bestellung und Vertrieb international: Nova MD GmbH, Vachendorf

ISBN 978-3-96443-725-9 (Print)

ISBN 978-3-906940-74-8 (E-Book)

© Editions Weblaw, Bern 2019

Alle Rechte sind dem Verlag Editions Weblaw vorbehalten, auch die des Nachdrucks von Auszügen oder einzelnen Beiträgen. Jede Verwertung ist ohne Zustimmung des Verlags unzulässig. Dies gilt insb. für Vervielfältigung, Übersetzung, Mikroverfilmung und die Einspeicherung und Verarbeitung in elektronischen Systemen.

Foreword

Hajime Yoshino has been a leading protagonist of legal logic and legal computing in Japan since the 1970s. Coming from the scholar community around Herbert Fiedler, Arthur Kaufmann, Ulrich Klug, Lothar Philipps, Jürgen Rödig, Ilmar Tammelo and Ota Weinberger, he has tirelessly advocated the application of logic in jurisprudence and thus also paved the way for legal computing. His mission, which he summarized in the «Logical Jurisprudence», is to further strengthen and expand the formal roots of jurisprudence in the transition field to legal computing.

Professor Hajime Yoshino is pleasant and humorous and it is a privilege to be able to meet him. He stimulates discussions and has an integrative effect.

In the course of designing this anthology it became clear once again that scientific practice is not only concerned with the application of formal logic as a core area, but that the word «formalization» describes a broader field which – assigned to the avantgarde – shapes the intellectual exploration of law and its environment through explicit structuring. The multitude of topics dealt with in this anthology is, to varying degrees of intensity, grouped around Yoshino’s preoccupations with jurisprudence, in which logic and formal thought are of fundamental importance.

We thank the jubilarian for his exemplary commitment to seeing law and logic as a unit and to shaping legal reality in accordance with this notion.

Ad multos annos!

Erich Schweighofer – Michał Araszkiewicz – Friedrich Lachmayer – Marijan Pavčnik

Vorwort

Hajime Yoshino ist seit den Siebzigerjahren ein maßgeblicher Protagonist der Rechtslogik und der Rechtsinformatik in Japan. Der Gelehrtencommunity um Herbert Fiedler, Arthur Kaufmann, Ulrich Klug, Lothar Philipps, Jürgen Rödig, Ilmar Tammelo und Ota Weinberger entstammend, hat er sich unermüdlich für die Anwendung der Logik in der Rechtswissenschaft eingesetzt und damit auch den Weg für die Rechtsinformatik mitbereitet. Sein Anliegen, das er in der «Logischen Jurisprudenz» zusammengefasst hat, ist Auftrag und Zeichen, die formalen Wurzeln der Rechtswissenschaft im Übergangsfeld zur Rechtsinformatik weiterhin zu verstärken und auszubauen.

Professor Hajime Yoshino ist eine angenehme und humorvolle Persönlichkeit. Es ist ein Privilegium, ihn treffen zu können. Er regt die Diskussionen an und wirkt integrativ.

Im Zuge der Gestaltung dieses Sammelbandes wurde erneut deutlich, dass es in der wissenschaftlichen Praxis nicht nur um den Kernbereich der Anwendung der formalen Logik geht, sondern dass das Wort «Formalisierung» ein weiteres Feld beschreibt, das – der Avantgarde zugeordnet – durch explizite Strukturierungen eine intellektuelle Durchdringung des Rechtes und seines Umfeldes aufbereitet. Die Vielzahl der in diesem Sammelband behandelten Themen gruppiert sich in unterschiedlicher Intensität um Yoshinos Anliegen einer Rechtswissenschaft, in welcher der Logik und dem formalen Denken ein grundlegender Stellenwert zukommt.

Wir danken dem Jubilar für seinen beispielgebenden Einsatz, das Recht und die Logik in einer Einheit zu sehen und die rechtliche Wirklichkeit in diesem Sinne zu gestalten.

Ad multos annos!

Erich Schweighofer – Michał Araszkiewicz – Friedrich Lachmayer – Marijan Pavčnik

Kevin D. Ashley – Legal Informatics and the CISG

Legal Informatics and the CISG

Kevin D. Ashley[1]

1.Introduction

Professor Hajime Yoshino has long taken an interest in The United Nations Convention on Contracts for the International Sale of Goods. The CISG (or the Vienna Convention) is a treaty that sets up a uniform international sales law. In the 1990s, he sponsored a series of international workshops on Legal Expert Systems for the CISG. During the intervening years, the CISG has been a centerpiece in his scholarly work in:

–  proposing a formal logical model of the CISG to assist with the problem of applying the treaty within and across legal systems, cultures, and languages,

–  illustrating the need for meta-reasoning about the validity of legal sentences, for modeling the Convention’s open-textured legal concepts using, for instance, fuzzy logic, and for logically modeling temporal reasoning,

–  demonstrating a legal expert system which, as it analyzes sales contract problems, reasons about the validity of legal sentences describing the parties’ obligations and rights and generates the changes in their legal relations,

–  using such a legal expert system as a pedagogical tool for teaching structured legal reasoning and a context for inculcating creative legal thinking.

In the meantime, more recent developments in the fields of legal informatics and AI and Law have focused on computationally modeling legal argument in more empirically descriptive ways and on applying techniques for text analytics and argument mining. These developments suggest how to improve legal information resources serving the community of CISG practitioners.

This paper examines the current technological approaches to modeling the CISG for purposes of legal information retrieval, relates them to the intellectual challenges Prof. Yoshino posed and the problems he addressed in his longtime focus on the CISG, and identifies areas for future work that might someday connect aspects of his models to the relevant legal texts.

The next section introduces the CISG, and section 3 surveys Professor Yoshino’s work with the Convention. Section 4 addresses some new research directions that could build on his introduction of and work on the CISG, followed by some conclusions in section 5.

2.Introducing the CISG to the AI and Law Community

According to Professor HARRY FLECHTNER (2009), a noted contract law scholar, «The United Nations Convention on Contracts for the International Sale of Goods (CISG) has been recognized as the most successful attempt to unify a broad area of commercial law at the international level.» It aims to increase international trade by providing a set of «substantive rules governing the rights and obligations of parties to international sales contracts.» With only eleven initial contracting states in 1988 when it first entered into force, the CISG has now been ratified by 89 states.

Among the CISG’s significant provisions, Flechtner includes those that address:

–  Interpretation of the parties’ agreement…;

–  The features, duration and revocability of offers;

–  The manner, timing and effectiveness of acceptances of offers;

–  The effect of attempts to add or change terms in an acceptance;

–  Modifications to international sales contracts;

–  The seller’s obligations with respect to the quality of the goods as well as the time and place for delivery;

–  The place and date for payment;

–  The buyer’s obligations to take delivery, to examine delivered goods, and to give notice of any claimed lack of conformity;

–  The buyer’s remedies for breach of contract by the seller…;

–  The seller’s remedies for breach of contract by the buyer…;

–  Passing of risk in the goods sold;

–  Anticipatory breach of contract…;

–  Exemption from liability for failure to perform, including force majeure;

–  Obligations to preserve goods that are to be sent or returned to the other party. (Flechtner 2009)

«The primary goal [of the drafters of the CISG] was to create uniformity in the rules for international sales, in order to supplant the complex and difficult-to-predict system that subjected international sales to the varying provisions of national sales laws.» (FLECHTNER 1998).

Professor Yoshino introduced the CISG to the field of Artificial Intelligence and Law in the early 1990’s in a series of four workshops and a number of publications. The CISG’s goal of a uniform international law of contracts may have been the reason why. Formalizing legal knowledge and inference in terms of logic has been a central theme of his career:

Typically, natural languages are used for expressing legal knowledge. Hence, if we want to formalize inferences with legal knowledge, we believe it would be better to use a formal language which has the ability to translate sentences expressed by a natural language correctly. (Yoshino 1997)

To this end, Professor Yoshino developed a Compound Predicate Formula (CPF) language and used it to translate CISG rules concerning the manner, timing and effectiveness of acceptances of offers into formal logical propositions that could be compiled.

One topic of discussion at the CISG Workshops was the relationship between a formal representation and uniformity of interpretation. This translation of the CISG’s rules into a formal language promised, among other things, to help achieve uniformity of interpretation of those rules, since, arguably, it could be assumed that the same formal propositions would underlie the various official language versions of the treaty. The formalization could serve as a kind of interlingua, which if it existed, would support uniformity in interpreting a treaty like the CISG. Alternatively, if the formal interlingua did not exist, it raised interesting questions of how and why the formal representations of the different language versions of the treaty would differ.

As noted, the CISG treaty was intended to promote uniformity in international sales law. Inevitably, as an international treaty, the CISG represents a compromise with multiple ambiguous formulations, vague terms and issues left for courts to address. Beyond that,

[t]he idea that the Convention comprises a set of uniform words--a single text describing a single set of rules equally applicable to all transactions within the scope of the CISG--turns out not to be strictly accurate. … [T}he Convention … is not … a single uniform text, but rather … a dizzying variety of texts. … [T]he very words comprising the Convention's rules will vary, often quite significantly, depending on: (1) the countries in which the parties to a CISG-governed transaction are located, and (2) the language of the tribunal resolving disputes about the transaction. This phenomenon results from the different language versions in which the Convention's sales rules are embodied, and the declarations or reservations made by various contracting states when they ratified the CISG. (FLECHTNER 1998, p. 189)

In addition, the treaty is not administered centrally. Multiple national courts have applied the treaty in its six official language versions (i.e., Arabic, Chinese, English, French, Russian, and Spanish) and according to their differing legal methodologies: civil law, common law, or using hybrid international interpretive techniques.

3.Professor Yoshino’s Work on the CISG

Professor Yoshino’s work on the CISG demonstrates some key themes of his AI and Law work: the goal of systematizing the law, the need for meta-reasoning about the validity of legal sentences, and the need for a legal expert system that performed such meta-reasoning. In (2011), Professor Yoshino laid out the logical structure of a system for reasoning about the duty relations that exist in a concrete legal problem and the legal rules that can be applied to decide it. In this framework, meta-sentences describe the validity of legal sentences; they determine if a legal rule is true and is applicable for solving the problem.

Problems involving the CISG provided a nice context for illustrating this meta-legal reasoning «from contracts through contract laws, constitutions and conventions up to basic legal rules.» (YOSHINO 2011) His papers illustrated detailed analyses of several variations of a contracts problem:

On April 1, a New York manufacturer, A (Anzai), dispatched to the Hamburg branch of a Japanese trading company, gives B (Bernard), a letter containing the following proposal: A will sell B a set of agricultural machines; the price of the machinery is $50,000; A will deliver the machinery to B by May 10; B must pay A the price of the machinery by May 20; the machinery will be transported by an American freight vessel. (2) The proposal reached B's letter box on April 8. (3) On April 9, B telephoned A and said, «I accept your offer». (4) A delivered the agricultural machinery to an American freight vessel at the port of New York on May 1. (5) The machinery was delivered to B's Hamburg branch on May 28. B examined the machinery on May 30. (6) B paid A $50'000 on May 31. (7) On August 10, the machinery malfunctioned because of a defective gear. (8) B notified A of the malfunction immediately. (9) On September 1, buyer B required seller A to repair the lack of conformity of goods by repair by October 1. (10) A did not repair the defect of the machine by October 1. (11) On October 10, B declared the contract void. (YOSHINO2012)

The goal of his legal expert system (LES) was to answer such questions as, «What kind of legal relations exist between Anzai and Bernard on the basis of the facts of the case at the following time points? April 5, April 15, May 15, June 5, August 15, September 15, October 5, November 15, and December 15.» For instance, it should be able to deduce that the proposition, «Anzai must repair the machine for Bernard,» is valid at time September 15th and to provide a proof in support of its answer.

3.1  Meta Legal Reasoning

Professor Yoshino proposed that these legal meta-rule sentences filled in certain important gaps in a code of law like the CISG. For instance, the «most fundamental of legal meta-rule sentences» stated that,

[r0] A legal sentence is valid at time T, if and only if the legal sentence becomes valid at time T1 before T & it is not the case that the sentence becomes null at time T2 before T.

Professor Yoshino states,

This legal meta-rule sentence is not found in any positive law code but implicitly presupposed in every law code. Whenever a legal sentence is applied, this rule sentence is to be applied to decide whether the legal sentence is valid. (2012)

Another legal meta-rule sentence fills in a gap in legal relations between conferring a right and imposing a duty:

[3aa2] A legal sentence «X has an obligation to do Z at time T1» becomes valid at time T, if Y exercises the right to require X to do Z at time T1 at time T & the legal sentence «Y has a right to require X to do Z at time T1» is valid at time T.

Thus, any time a substantive legal rule in the CISG was applied, such as

[Article 46 (3)] If the goods do not conform with the contract, the buyer may require the seller to remedy the lack of conformity by repair….,

r0 would be invoked along with some helper legal meta-rule sentences to prove its validity as part of a CISG which is valid as a whole, and [3aa2] would help prove the validity of the seller’s obligation on September 15 to repair the machine.

3.2  Temporal Legal Reasoning

The references to time variables in these two legal meta-rule sentences suggest why Professor Yoshino pursued AI techniques for modeling temporal reasoning. (VILA & YOSHINO 1998) presents LTR, a time logic adapted to some of the requirements of legal reasoning. Temporal incidence predicates (TIPs) were defined such as holds, occurs, holds_​at, holds_​on, and applied to represent various provisions of the CISG dealing with offer and acceptance. Given a formalized representation of the facts in a problem like the Anzai scenario, the LTR formalization could answer a query, «Is the contract concluded?» as «YES, as of April 15» by generating a map-like representation of the events and provide an explanation in terms of the rules that fired. LTR did not address tasks that involve meta-reasoning about the validity of statutes and laws over time, which was the focus of future work in AI and Law. See, e.g. (PALMIRANI, 2011). Professor Yoshino’s work on CISG is one of the few bodies of work to address modeling temporal reasoning in law.

3.3  How Complete a Formalization is Enough?

Despite the complexity in Professor Yoshino’s examples of the proofs in which meta legal reasoning establishes the validity of a substantive rule of the CISG from the validity of the Convention as a whole, it is interesting to consider what they do not address. The examples did not address all the issues that may arise concerning whether the CISG applies to the transaction in question, including some issues that are more likely to be contested than the validity of the Convention as a whole. For instance, according to Article 1(1)(a), the CISG:

applies to contracts of sale of goods between parties whose places of business are in different States: (a) when the States are Contracting States; or (b) when the rules of private international law lead to the application of the law of a Contracting State.

Whether the States are Contracting States for purposes of this provision raises issues about whether they have elected to be bound to certain provisions. Moreover,

even where all requirements for the applicability of the CISG are met, the CISG does not necessarily have to apply….» [A]rticle 6 allows the parties to opt-out of the CISG or portions of it …. The exclusion can be express or implied …, total or partial. To exclude the Convention ex article 6, there must be an agreement between the parties …. Article 8 identifies interpretive criteria to be taken into account to determine whether there is such an agreement. Article 9 lays down rules on trade usages and course of dealing, which must also be taken into account to determine whether the parties intended to exclude the CISG. (FERRARI 1996)

This is not to criticize Professor Yoshino for oversimplifying or leaving out some necessary consideration. Surely, he would jump at the chance to apply his formal methods to any issues of validity raised by each of these additional provisions. Rather, it is to question whether he has attempted to include too much, that is, to question the practicality of his goal of developing a formal systematization of law, at least to the extent it aims to be complete. While the validity of any and every provision could be at issue, frequently it is not contested; the advocates are presumed to accept the validity of legal authorities like the CISG unless they expressly contest it. Presumably, a system should not generate proofs of every possible issue but focus on those issues that are contested or those, which, given the facts of the problem, can reasonably be contested and for which a system could construct reasonable arguments.

3.4  Case-based Reasoning and Argument

Two trends in AI and Law addressed this kind of problem of selection, case-based reasoning and the development of computational models of legal argument. Historically, they developed in that order, but today, CBR is treated as one in a number of components of a computational model of legal argument, as is logical deduction from a legal rule. The need to make or respond to an argument leads a program via critical questions or other control mechanisms to discover the reasonable grounds presented in the fact situation for developing a response. See (ASHLEY 2017, Ch. 5).

Professor Yoshino’s work applying fuzzy logic to the CISG reflected both of these trends. The Fuzzy Legal Expert System, FLES, was intended to make case-based arguments concerning open-textured terms in the CISG such as «sufficiently definite,» as in «A proposal is sufficiently definite if it indicates the goods and expressly or implicitly fixes or makes provision for determining the quantity and the price.» Article 14. (HIROTA, ET AL. 2000) Professor Yoshino and his co-authors had developed a case-base of eight CISG decisions and factors concerning prices for goods such as whether there is a market or commercial substitute for a component. (MINGQIANG, ET AL. 1998) Factors are stereotypical patterns of fact that tend to strengthen or weaken a side’s argument in support of a claim or issue. See (ASHLEY 2017, Ch. 3) FLES invoked fuzzy logic to assess the similarity of cases to a current fact situation and supported and responded to claims resolving the ambiguous term’s application.

3.5  Educating Creative Legal Minds

FLES was intended to serve as an educational system for teaching law students argumentation skills. Its predecessor legal expert systems, especially LES-5, also played a role in legal education, an important focus of Yoshino’s work on the CISG.

The input to the program is a formalization of the facts of the Anzai problem above. The program outputs a kind of time map of the legal relations among the buyer (Bernard) and seller (Anzai) in the above problem at various points of time during which a contract becomes effective, the seller is obligated to deliver the goods to buyer by May 10, seller delivers the goods, the buyer is obligated to pay the price to seller by May 20, trouble occurs, the buyer may require seller to remedy the lack of conformity of the goods by repair, the buyer exercises a claim to repair, etc. At each point the program specifies the correlative rights and obligations of the parties and can provide a justification in terms of the legal rules of the CISG and the legal meta-rule sentences like [3aa2] above.

When I first saw the output of LES-5 (or one of its predecessors) I was struck with the way in which it visually indicated the various successive legal states of affairs associated with the life of a particular sales contract under the CISG. The utility of such a system for teaching law students about contract law was immediately apparent. The program’s outputs are a window through which students can perceive not only the CISG’s substantive provisions of contract law such as offer and acceptance but also the often unstated supplemental rules of statutory interpretation, the legal meta-rule sentences, that, in Professor Yoshino’s view, are an all-important component of a formalization of legal reasoning, and thus a necessary element of understanding legal reasoning. The gaps in legal reasoning, the presumed knowledge that might otherwise elude a student’s attention, are thus made explicit.

Formalization and systematization are key in Professor Yoshino’s thought on educating creative legal minds.

Deductive reasoning in law is not a given system. To systematize the law is not to recognize a given system, but to create a system. People who attempt to understand the law have to create it themselves. Actual legal regulations in themselves do not compose a strict system. People who apply and interpret the law systematize it themselves. The systematization of law is creative legal thought. (YOSHINO & SAKURAI 2005)

LES-5 provided students examples of a systematization of parts of the CISG. Then, through a series of problems in which components of the systematization were intentionally omitted, students were invited to come up with their own components to fill in the gaps. After all, filling in the gaps of a law code with legal meta-rule sentences was the way in which Professor Yoshino’s creative legal mind expressed itself, and it is natural that he would seek to enable his law students to learn to do the same. Query whether any AI and Law scholar from a civil law jurisdiction has done more than Professor Yoshino to model the kinds of interpretive rules that play such an important role in civil law reasoning.

4.Future Research on the CISG

Having briefly surveyed the arc of Professor Yoshino’s CISG research, it is interesting to consider the kinds of research challenges this convention may still hold for a new generation of AI and Law researchers. Legal expert systems are still an important component of AI and Law, but they are used primarily for document assembly, business compliance, and directing clients to the right human experts for their problems. As noted, recent developments in the fields of legal informatics and AI and Law have focused on applying text analytics and argument mining to extract semantic information from legal texts and on modeling legal arguments in more empirically descriptive ways.

4.1  Applying Legal Text Analytics to the CISG

Text analytics or text mining involves discovering knowledge using linguistic, statistical, and machine learning techniques to model some information content in text archives. Legal text analytics involves archives of legal texts including regulatory texts (e.g., statutes, treaties, and regulations), case decisions, and contracts. Argument mining involves the text analytic discovery of argument-related information. Machine learning (ML) refers to programs that use statistical means to induce or «learn» models from data. The resulting models can predict outcome classifications for a new instance or case. See (ASHLEY 2017).

Today, there are more than 3,000 cases involving the CISG and 10,000 case annotations.[2] A CISG corpus of electronic versions of the texts of these cases, in the original language and English translations, is readily available from public sources. While the Pace Law School’s CISG Database is an excellent resource, new approaches to legal information retrieval could dramatically improve the accessibility of these materials beyond the use of keyword searches and indexing.

Frequently in CISG cases, a complaining party brings suit against a responding party for breach of contract, and either succeeds in showing that the facts comply with the CISG’s requirements for recovery or not. When a number of court decisions focus on the same CISG provisions and present factually similar scenarios, the courts sometimes reach inconsistent results, based on inconsistent interpretations of the CISG. An interesting challenge for AI and Law researchers would be to detect such inconsistencies in treaty interpretation automatically.

An initial step would be to improve legal information retrieval resources serving the community of CISG practitioners by enabling them automatically to identify and retrieve cases that deal with particular issues, arguments, and results. One could assemble a corpus of CISG case texts and develop a program to determine the legal requirements at issue in a case, the court’s conclusions regarding the issues, the types of arguments employed, and whether other courts have addressed the same issues in factually similar circumstances but reached different outcomes (or the same result but for different reasons). The main hypothesis to test would be that a program can detect in the corpus of CISG texts, with acceptable levels of classification accuracy and confidence, the legal requirements, conclusions, argument types, and instances of differing legal results or treaty interpretations.

An important step in this process would be to develop a rule tree, that is a tree of authoritative rule conditions for applying the CISG’s substantive contract provisions. (WALKER, ET AL. 2011) This process would be similar to what Professor Yoshino and his colleagues have already done: manually analyzing the statutory and treaty provisions in order to encode the legal rules. (see, e.g., YOSHINO 1997) An interesting question will be the extent to which the rule tree would need to include legal meta-rules of the types Professor Yoshino identified. The rule tree is not intended to serve as a support for deductive reasoning but as an index into the legal issues actually addressed in the decisions. In studying the CISG cases, citation networks could identify the most influential precedents and one could examine these to identify legal meta-rules that the courts have expressly applied.

Then, the goal would be to find ways in which a program could automatically detect those legal issues that a decision addresses and where they connect to the rule tree. Citations will also play a role in this detection, but machine learning and rules for detecting standard formulations and functional abstractions of legal provisions will also probably assist. The goal is to annotate the case texts in syntactic and domain general semantic terms and in terms of the evidential reasoning chains from facts to legal conclusions (WALKER, et al. 2011) and of factors and other functional and magnitude/​certainty indicators. Professor Yoshino’s work on FLES (MINGQIANG, ET AL. 1998; HIROTA, ET AL. 2000) is one indication that factors can be useful in the CISG domain on such issues as whether quantity and price are «sufficiently definite».

Since they deal with contracts and contract law, CISG cases present an opportunity to examine the contractual provisions and arguments that courts have considered in applying the CISG and to attempt to model them computationally, something that has not been attempted in AI and Law since the 1990s. Today, text analytic advances in analyzing contracts may enable researchers to gain some traction on modeling such arguments, for example, where courts interpret contractual language attempting to opt out of the CISG.

Implementing and integrating these kinds of annotations would require defining a formal system of layered annotation types using, for example, the UIMA framework.[3] The type system, a kind of ontology for text analysis of legal decisions and arguments concerning the CISG, would define the required annotations to represent terms (including functional terms from the CISG) syntax, semantics, legal argumentation and evidential reasoning and how these layered annotations are linked together as the different tools annotate a case text.[4] The formal type system would be used to combine human annotation (a gold standard) with the outputs of automatic components, such as pattern matching with heuristic rules or statistically-trained language models. LUIMA, a legal unstructured information management architecture and tools, adapt UIMA to some of the requirements of a legal domain. (GRABMAIR 2011).

The program’s annotations could be evaluated against a ground truth: the legal cases provide an objective record of who won, and often, for each issue, one can objectively determine who won that issue, why, or why not. In cross validation experiments, a standard procedure for evaluating a machine learning program, these cases provide a basis for objectively measuring performance and effectiveness in detection. See (ASHLEY 2017, Ch. 10) Since examples of inconsistent judicial decisions concerning the CISG attract legal scholarly interest, the secondary literature also provides a basis for identifying cases with inconsistent interpretations.

4.2  The Multi-language CISG

A continuing focus on CISG research could also present an opportunity to address issues of translation in AI and Law. The original texts of both the CISG provisions and cases have been published in languages other than English. Google Translate can translate texts of CISG provisions from one language to another. Consider the example below of French and English versions of Article 46(3). The example shows a translation of the official French version of the above-cited Article 46(3) into English (e) and of the official English version into French (c) and back into English (d). Comparison of the English texts (a), (d), and (e) demonstrates serviceable translations but some concepts appear to wander. A «request for repair» in (a) has become a «claim for compensation» in (d) and «The remedy» in (e). While a human reader of (e) may recognize that «The remedy» refers to the request for repair in the previous sentence, a «request for repair» has a different meaning than a «claim for compensation». In addition, the concept of «notice» in (a) and (d) becomes a «denunciation» in (e).

Article 46(3)

(a) Official English Text: If the goods do not conform with the contract, the buyer may require the seller to remedy the lack of conformity by repair, unless this is unreasonable having regard to all the circumstances. A request for repair must be made either in conjunction with notice given under article 39 or within a reasonable time thereafter.

(b) Official French Text: Si les marchandises ne sont pas conformes au contrat, l'acheteur peut exiger du vendeur qu'il répare le défaut de conformité, à moins que cela ne soit déraisonnable compte tenu de toutes les circonstances. La réparation doit être demandée au moment de la dénonciation du défaut de conformité faite conformément à l'article 39 ou dans un délai raisonnable à compte de cette dénonciation.

(c) Google Translate Version of (a) into French: Si les marchandises ne sont pas conformes au contrat, l'acheteur peut exiger du vendeur qu'il remédie au défaut de conformité par la réparation, sauf si cela est déraisonnable compte tenu de toutes les circonstances. Une demande de réparation doit être faite conjointement avec un avis donné en vertu de l'article 39 ou dans un délai raisonnable par la suite.

(d) Google Translate Version of (c) back into English: If the goods are not in conformity with the contract, the buyer may require the seller to remedy the lack of conformity by the repair, unless this is unreasonable in the light of all the circumstances. A claim for compensation must be made in conjunction with a notice given under section 39 or within a reasonable time thereafter.

(e) Google Translate Version of (b) into English: If the goods are not in conformity with the contract, the buyer may require the seller to repair the lack of conformity, unless this is unreasonable having regard to all the circumstances. The remedy must be requested at the time of the denunciation of the lack of conformity made in accordance with Article 39 or within a reasonable time after the denunciation.

Recently, there has been interest in an interlingua of universal «concepts» that may emerge from Google’s multilingual Neural Machine Translation (NMT) (JOHNSON, ET AL. 2016). Although this interlingua is quite different from Professor Yoshino’s suggestion of a uniform logical formulation underlying the CISG and its language versions, it begs the question of whether translation of legal treaties requires applying more knowledge of the meaning of legal terms, a question that his CISG Workshops raised.

5.Conclusions

In introducing the field of AI and Law to the CISG, The United Nations Convention on Contracts for the International Sale of Goods, Professor Hajime Yoshino substantially enriched the field’s domain of discourse for many years. His long-standing interest in the Convention produced formal logical models of the CISG and legal expert systems that demonstrate the need for meta-reasoning to interpret the validity and meaning of a legal code and that introduce law students to creative legal thinking by showing them how to develop such formalizations.

This Festschrift in Professor Yoshino’s honor should be an occasion for the field to consider how more recent developments in the fields of legal informatics and AI and Law, focusing on computationally modeling legal arguments in actual cases and on applying techniques for text analytics and argument mining, can improve legal information resources serving the community of CISG practitioners. These developments may someday connect aspects of his models to the actual legal texts of the treaty and the cases interpreting it.

6.References

ASHLEY, K. D. (2017). Artificial intelligence and legal analytics: new tools for law practice in the digital age. Cambridge University Press.

Ferrari, F. (ed.) (1991). Cross-References and Editorial Analysis, Article 1, http://​www.​cisg.​law.pace.edu/​cisg/​text/​cross/​cross-1.html (visited 02 August 2018).

FLECHTNER, H. (1998). The Several Texts of the CISG in a Decentralized System: Observations on Translations, Reservations and other Challenges to the Uniformity Principle in Article 7(1), 17 Journal of Law and Commerce (1998) 187–217

FLECHTNER, H. (2009). Introductory Note, United Nations Convention on Contracts for the International Sale of Goods http://​legal.un.org/​avl/​ha/​ccisg/​ccisg.html

GRABMAIR, M., ASHLEY, K., CHEN, R., SURESHKUMAR, P., WANG, C., NYBERG, E., and WALKER, V. (2015). Introducing LUIMA: An Experiment in Legal Conceptual Retrieval of Vaccine Injury Decisions using a UIMA Type System and Tools. Proceedings of the 15th International Conference on Artificial Intelligence and Law. ICAIL 2015. pp. 1–10, New York, NY, USA: ACM.

HIROTA, K., TAKAMA, Y., XU, M., & YOSHINO, H. (2000). Legal argument in fuzzy legal expert system (FLES). In Industrial Electronics Society, 2000. IECON 2000. 26th Annual Conference of the IEEE (Vol. 1, pp. 207–210). IEEE.

JOHNSON, M., SCHUSTER, M., LE, Q. V., KRIKUN, M., WU, Y., CHEN, Z., ... & HUGHES, M. (2016). Google's multilingual neural machine translation system: enabling zero-shot translation. arXiv preprint arXiv:1611.04558.

MINGQIANG, X., HIROTA, K., & YOSHINO, H. (1998). A Fuzzy Theoretical Approach to Representation and Inference of Cases in CISG. Artificial Intelligence and Law.

PALMIRANI, M. (2011). Legislative Change Management with Akoma-Ntoso. Sartor, G., Palmirani, M., Francesconi, E., and Biasiotti, M. (eds), Legislative XML for the Semantic Web. Law, Governance and Technology Series, vol. 4. pp. 101–130. Springer Netherlands.

VILA, L., & YOSHINO, H. (1998). Time in automated legal reasoning. Information and Communications Technology Law, 7(3), 173–197.

WALKER, V., CARIE, N., DEWITT, C, and LESH, E. (2011). A framework for the extraction and modeling of fact-finding reasoning from legal decisions: lessons from the Vaccine/​Injury Project Corpus. Artificial Intelligence and Law, 19(4), 291–331.

YOSHINO, H. (1997). On the logical foundations of compound predicate formulae for legal knowledge representation. Artificial Intelligence and Law, 5(1–2), 77–96.

YOSHINO, H. (2000). Use of Legal Expert System LES-5 in Legal Education.

YOSHINO, H. (2011). The systematization of law in terms of the validity. In Proceedings of the 13th International Conference on Artificial Intelligence and Law (pp. 121–125). ACM.

YOSHINO, H. (2012) The Logical Analysis of the Concept of a Right in Terms of Legal Meta-Sentences, in: Jusletter IT 29 February 2012.

YOSHINO, H., & SAKURAI, S. (2005). A knowledge-based systems approach to educating creative legal minds. AILE.

Katie Atkinson / Trevor Bench-Capon – The Roles of Dimensions and Values in Legal CBR

The Roles of Dimensions and Values in Legal CBR

Katie Atkinson / Trevor Bench-Capon[5]

Keywords: legal case based reasoning, dimensions, factors, values

Abstract: In this paper we build on two recent attempts to formalise reasoning with dimensions. Both of these approaches effectively map dimensions into factors, which enables propositional reasoning, but we show that sometimes a balance between dimensions needs to be struck. To permit trade-offs we need to keep the magnitudes and reason more geometrically. We also revisit the link between dimensions and values, arguing that values play a number of distinct roles, not only explaining preferences between factors, but also ensuring that all the purposes of the underlying law are considered.

1.Introduction

One of the major tasks addressed by AI and Law over the last three decades or so has been to come to a good understanding of the representation of, and reasoning with, legal cases. This has also been central to the work of Professor Yoshino. In 1984 he organised the Legal Expert System Association (LESA) in Japan [36], and began to develop legal expert systems, (LES and LES-2 [35] in the domain of contract law). LES-2 was largely within the logic programming paradigm of [30]. Further developments included the language CPF (Compound Predicate Formula) [32], specifically to represent legal knowledge, and methods to allow meta inference [33]. Professor Yoshino continued to work on contracts, especially in the context of the Convention on the International Sale of Goods (he organised a series of workshops on this topic collocated with ICAIL in the 90s) and he came to recognise the need for case law, and the need to use fuzziness to handle this [20]. This paper describes our own current work on this topic, in particular how best to handle the problem of non-Boolean features of cases, necessary to capture certain kinds of reasoning with legal cases.

Our work is developed within the tradition of HYPO and its successors, a line of research described in [10], which has only recently started to again pay serious attention to non-Boolean features. Although such features were present as dimensions in HYPO, they have (with occasional exceptions such as [12]) tended to be simplified to Boolean factors in the drive to obtain a good understanding of the logic of precedential reasoning. Recently, however, as described below, interest in such features has undergone a significant revival.

As told in [10], the story begins with the dimension-based HYPO [7] and moves through the factor-based CATO [5], to expression as a set of rules [25] enabling a formalisation of factor-based reasoning by Horty in [22], which was refined by Rigoni in [27]. Factors can be seen as stereotypical patterns of facts, either present or absent in a case, and, if present, favouring either the plaintiff or the defendant. Dimensions, in contrast, are ranges of values (either numeric or enumerated), running from an extreme pro-plaintiff point to an extreme pro-defendant point. The applicability of dimensions to a case, and the point at which the case lies, is determined by the case facts, and the dimension may favour either party. The relationship between dimensions and factors is discussed in detail in [29]. Since the mid-90s, factor-based representations have been the main focus, and, although dimensions have always had their advocates [12], it is only more recently that dimensions have been revived as a way of connecting factors to the facts, and providing a way of capturing additional nuance (e.g. [26], [2] and [4]). More recently Horty has attempted to extend his formalism to accommodate dimensions [21], and in [28] Rigoni has critiqued this approach from the standpoint developed in [27]. In parallel with these developments, there has been exploration of the relationship between case law decisions and the purposes, or social values, they promote. The idea of its associated purpose as a measure of strength of a dimension was introduced in [14] and most fully expressed (in terms of values) in [13]. Recent discussions of the role of values, considering dimensions as well as factors, can be found in [1] and [2].

In this paper we will consider the role of values and their relation to dimensions in the light of [21] and [28]. After some background, we will consider how to argue with dimensions in legal CBR, and the role of values. We do so using the domain knowledge represented as an Abstract Dialectical Framework (ADF) [16] as described in [3]. In particular we will show that:

–  Any legal CBR problem can be reduced to a series of steps involving at most two dimensions, so that higher dimensional spaces need not be considered;

–  The non-leaf nodes of the ADF can be seen as being one of five kinds, as determined by the types of their children;

–  For some nodes, dimensions cannot be reduced to factors and need to retain their magnitude, to permit trade-offs;

–  Values are required to play several different roles, not just the expression of preferences between rules as in [13].

2.Formalising Factors and Dimensions

The formalisations of factor-based reasoning of both Horty and Rigoni are based on the method of expressing precedents as rules found in [25]. In that paper a case is considered to be a triple (P, D, o), where P is the set of all pro-plaintiff factors present in the case, D is the set of all pro-defendant factors present in the case and o is the outcome, either plaintiff (π) or defendant (δ). Now P → π will be the strongest reason to find for the plaintiff and D → δ will be the strongest reason to find for the defendant. We can therefore deduce that either P ≻ D or D ≻ P depending on the value of o. These preferences permit only a fortiori reasoning. Although [25] has a notion of rule broadening as a dialogue move, a key insight of Horty is that P → π may be stronger than is required and some subset of P may be sufficient to defeat D. Horty does not specify how exactly the subset is determined, but it could be interpreted as the ratio decidendi of the case. In general the use of P gives rise to what Horty terms the rule or result model and attributes to Alexander [6], and the subset what he terms the reason model, and attributes to Lamond [23].

Figure 1. Dimension after 2 cases

Horty’s key idea in [21] is that dimensions can be mapped into factors, with the position on the dimension occupied by a particular case determining (through precedent cases) whether the corresponding factors are present or absent in that case. Note that the point at which the factor becomes present may not be the point on the dimension in the case facts of the precedent, which would be the result model. Instead the factor might become present at a point weaker for the side it favours so including the precedent and some other future cases, giving the reason model. In his example (taken from [25]), a person is attempting to show a change of fiscal domicile on the basis of absence from his home country. In the example a person is absent abroad for 36 months and change is found on the grounds that the absence is longer than a year. Thus the factor is present on the results model if the absence is at least 36 months and on the reason model if the absence is greater than a year. On his approach, however, Horty finds that, when using magnitudes, the result and the reason models collapse into one. Moreover, the reason does not constrain. Horty considers a second case in which the person has been absent for 18 months, but the court wishes to apply a stricter standard and rule against change on the grounds that it was less than two years. Horty wishes to say that the court can consistently decide in this way, although it cannot offer as a reason that the absence was below a threshold greater than three years, which would contradict the result of the precedent case. Rigoni objects in [28] both to the collapse of the two models, and to allowing the court to decide that 18 months is not enough for change, which he claims is counter intuitive, given the reason expressed in the precedent. Essentially Rigoni is happy to adopt the reason given in the decision, and disregard the particular instantiation using the facts of the case in which the reason was stated. That might be expressed as Rigoni treating the reason as ratio, whereas Horty treats it as obiter.

The dimension and the two cases in the example are shown in Figure 1. There are a number of points of interest within this dimension. One set is the positions occupied by the precedent cases. Another set is the positions used to express reasons (if any) in the precedent cases. Finally we have the point, identified in [28] but not explicitly in [21], at which the dimension ceases to favour no change and begins to favour change. Rigoni terms this the switching point (SP). The question is where SP lies. For Horty it lies somewhere to the right of 36, whereas for Rigoni it cannot lie to the right of 12, if Case 1 is serving as a precedent. Rigoni then presents an alternative way of formalising dimensions which avoids the collapse of the two models and satisfies his intuition to disallow the decision in Case 2. Rigoni suggests that a dimension should be thought of as a series of factors and uses magnitude to constrain their relative strength and SP to determine their polarity. This is essentially also the approach of [26] and [11], although presented less formally in those papers.

Here we will avoid reducing dimensions entirely to Boolean factors, which is what enables the reasoning to be expressed in terms of propositions and rules, convenient as this is for arguing in a conventional style. We want to avoid this reduction because it excludes the notions of trade-offs and balance between dimensions, used in some legal decisions [24]. Instead we will build on the ideas of [9] and look at the possibilities of a different flavour of argumentation, based on geometry rather than rules. A further aim of the paper is to relate the dimensions to values, which are not discussed in either [21] or [28].

3.Arguing with Dimensions

From Figure 1, we can see that the dimension can be divided into a number of zones. Case1 argued that 12 months should suffice for change, and that 36 months conclusively established change. Thus less than 12 months should presumptively find for no change (no precedents to establish a conclusive minimum absence have yet been decided), between 12 and 36 months there is a presumption for change and after 36 months change is conclusively established. The reason is not, however, followed in Case2, which establishes less than 18 months as conclusive for no change, between 18 and 24 months as presumptively favouring no change, reducing the range presumptively favouring change to 24 to 36 months. Beyond 36 months remains conclusive for change. This would put the SP somewhere between 18 and 36 months, presumptively, based on these two cases, at 24 months.

Given such a picture we would expect all the cases for change to appear to the left of SP and all the cases for no change to appear to the right of SP. But it is unlikely that this will be so, because it is unlikely that absence will be the only dimension to be considered. Typically a case can be distinguished by pointing to a different aspect which is more favorable to the other party. In [21] Horty adds a second dimension to the example, namely the percentage of income earned abroad. The idea is that the greater this percentage, the stronger the case for change, and so it may well be that there is a kind of trade-off whereby a larger percentage of income from abroad will require a shorter absence. If only the absence dimension is considered, some cases decided for change may appear to the right of cases decided for no change. This may be explained by looking at another dimension: perhaps the percentage of income is greater in those cases. The arguments resulting from considering two dimensions were considered in [9].

Figure 2. Case 3 with 2 dimensions, showing alternative decisions

In two dimensions we need to think in terms of north-west and south-east of the various points, rather than left and right. The facts of the case and its result define an area where the decision must be followed (black in Figure 2), and the reason given offers a hypothetical set of facts that creates an area which presumptively favours the winning side (grey in Figure 2). A new case may then fall into an area not yet covered by precedents and depending on the outcome, it will claim some of the space for the winning side. Figure 2 illustrates the situation, for the domicile example using absence and income percentage and a third case.

4.Use of 2-Regular ADF

In [9] the discussion was always in terms of two dimensions, but it was left open as to whether higher dimensional spaces might require consideration. In fact, just as any set of relations can be expressed in terms of binary relations and any k-SAT problem can be expressed as 3-SAT, it is possible to represent any domain so as to ensure that no more than two dimensional spaces are needed. In [3] the ANGELIC methodology for representing domain knowledge as an Abstract Dialectical Framework (ADF) [16] was presented. Formally an ADF forms a three tuple: a set of nodes, a set of directed links joining pairs of nodes (a parent and its children), and a set of acceptance conditions, one per node. The nodes represent statements which, in this context, relate to issues, intermediate factors and base level factors[6]. The links show which nodes are used to determine the acceptability of other nodes, so that the acceptability of a parent node is determined by its children. The acceptance condition for a node states how precisely its children relate to that node. In [1] it was shown that such an ADF could be rewritten as a 2-regular ADF, in which every non-leaf node has at most two children. Since the acceptability of a node in an ADF depends only on its children, this means that we need never consider more than two dimensions to resolve the acceptability of a node, and, since an ADF produced by the ANGELIC methodology forms a tree, the topmost node can be resolved without the need to consider more than two nodes at any given step.

Also at this point we should note that not every aspect of a case requires representation of magnitude. In the original HYPO [7] in 10 of the 13 dimensions only the two extreme points were of interest. Such aspects are better seen as two factors, and always favouring a particular side. Recognising this, in [28], Rigoni describes cases using both factors and dimensions. Even when the aspect has genuine magnitude, in the context of the particular domain it may be unnecessary to go beyond factors. Thus age is clearly an aspect with magnitude, but in a particular context, it may be possible to reduce it to a set of factors using thresholds. For example it may be that age is relevant only to distinguish between adults and minors, so any age above 18 is equivalent to every other age above 18. (Note both that domain case law may not be required to fix the threshold and that the threshold may change: pre-1970, the UK threshold for the age of majority was 21 rather than 18.) This means that any given node in our 2-regular ADF (other than non-leaf nodes which are instantiated directly from the case facts) may have as children any of the following five combinations:

1. two factors;

2. one dimension and one factor;

3. two dimensions;

4. one factor (the other child is a dummy node, for example, true);

5. one dimension (the other child is a dummy node, for example, true).

(1) is found in factor-based reasoning as formalised in [22] and [27].

In (2) the factor provides a context for the consideration of a dimension. Suppose that in the fiscal domicile example, citizenship is a factor to be considered: if the person is a UK citizen, a longer absence may be required before a change is made. Note that this aspect has no natural interpretation with magnitude: either one is a UK citizen or not. The SP for the child with magnitude will depend on the context supplied by the factor child.

In (3) we have the kind of trade-off illustrated in Figure 2. The two dimensions describe points in a two dimensional space, and a line is drawn separating the area favouring one outcome from the area favouring the other outcome. There is no difference between dimensions which are best thought of as continuous and those best thought of as a set of discrete points: in the latter case the space can be represented as a lattice, but this still needs to be partitioned into the areas favouring the two outcomes.

Examples of (4) should be rare: the child can simply replace the parent. Finally in (5) we have a way of implementing thresholds. Thus the parent will be something like sufficient absence, and the purpose of the node is to provide a means of converting a dimension into a factor, much as envisaged by Horty in [22]. Similarly a set of such nodes, all with the actual point of the dimension as their child, would produce the set of factors envisaged in [28] and [26].

Parent

Child 1

Child 2

Acceptance

Trade Secret

Misappropiation

Info Trade Secret

Info Misappropriated

AND

Info Trade Secret

Information Valuable

Maintain Secrecy

AND

Info Misappropriated

Not OK Means

Improper Means

OR

Not OK Means

Info Used

Confidential Relationship

AND

Table 1. IBP Logical Model as 2-Regular ADF

5.Relation with Values

Now we can reintroduce a relationship with purposes or values. The idea of values derives from [14] in which values were used to explain preferences between competing factors, and hence to resolve conflicts for which there was no precedent involving precisely these factors. In the absence of an exact precedent in terms of factors, if the factors involved relate to values between which a preference had been expressed in a precedent, that value preference can be applied to determine which factor should be preferred. This allows us to go beyond a fortiori reasoning. This idea was the basis for the formalisation of theory construction found in [13]. As the exploration of values developed it became recognised that, since the object of law is, as argued in [14], to fulfill certain purposes, the various aspects of the case that need to be considered (that is, the factors and dimensions) are those required to fulfill these purposes, to ensure the promotion of, and avoid the demotion of, certain values. Thus the existence of factors and dimensions in case law domains was justified by their role in enabling the consideration of particular values. This was the basis of the representation of cases in [18] in which there was a one-to-one relation between dimensions and values, and in [11], although there some values were represented by two distinct dimensions. In [39] it was recognised that values might play two roles: justifying the presence of a rule, or justifying the inclusion of a particular antecedent in a rule. We will begin by illustrating the ideas with the logical model of the US Trade Secrets domain used in [17], shown as a 2-regular ADF in Table 1.

As given in [18] and [11], US Trade Secrets requires consideration of 5 values:

–  Reasonable Efforts (RE), requiring the plaintiff to take reasonable steps to protect the information,

–  Material Worth (MW), requiring that the information have value,

–  Questionable Means (QM), requiring the defendant to refrain from criminal or dubious conduct,

–  Legitimate Means (LM), allowing the defendant to acquire the information independently (that is without using the information), by research, etc., and

–  Confidential Relationship (CR), enforcing respect to confidentiality agreements.

These values are represented by the leaf nodes of the logical model. Where the children are linked by AND, we ensure that both values are promoted, and where they are linked by OR we ensure that at least one of the values is promoted. Thus the role of nodes with two factors as children linked by AND or OR is to ensure that required values are given their due consideration. But there are also cases where the polarity of the two children is different: effectively the connective is UNLESS. An example concerning Trade Secrets taken from [1] is Info Valuable which has the pro-child Info Useful and the con-child Info Available, i.e. the information is valuable if it is useful unless it is available elsewhere. Here we can express a preference between values: the value of the exception is preferred (since otherwise the exception would have no effect). Here we can say that the value of discovery by Legitimate Means outweighs the value of Material Worth. No matter how valuable the information, it cannot be protected against legitimate discovery by a competitor. Unlike [13], different preferences may be used in different nodes.

The second kind of node is where we have a factor providing a context for a dimension. This can be illustrated using our fiscal domicile example. Absence is there to promote stability. But the length of absence might be considered differently for different types of citizen. Thus UK citizens might require a longer absence than citizens of other countries who had been working here on a long term, but not permanent, posting. Thus we may envisage a parent sufficient given citizenship, with children UK citizen and absence. What we have here is two distinct dimensions of the sort shown in Figure 1. The cases that fall on each dimension, and the switching point, will depend on the value of the citizenship factor in the particular cases. The value served here is stability, but the context allows consideration of the value of mobility of labour, since we are allowing non-UK citizens an easier path to restoring their original fiscal domicile. Thus we are able to consider two values, or to consider what promotes a value in a particular context. Similarly nodes of type (5) allow consideration of what is sufficient to promote a value, but here no context need be considered: the switching point at which the dimension becomes sufficient is the same for all cases. This permits a threshold for a factor to be determined by precedents, as envisaged in [21], and argument about which threshold is appropriate.