1.
Introduction ^
The main ideas of the PPSS have been already presented in [Araszkiewicz, Łopatkiewicz, Zienkiewicz 2013a and 2013b]. The objective of this paper is more specific. The paper focuses on the structure of the system’s knowledge base and inference engine with particular emphasis on Case-Based Reasoning (CBR) structures, that is, on the role played by dimensions and factors in the system and on the operation of rule-based inference engine on the information provided by CBR knowledge representation structures.
2.
The Background of the PPSS Project ^
As for the character of the dispute that may arise between the parties to the negotiated agreement it is necessary to emphasize that due to the fact that the divorce proceedings are adversary, the parents may be naturally inclined to adopt an adversarial attitude also as regards the work on the parent plan. However, as the main premise concerning acceptability of their agreement is its compatibility with the well-being of the child, the optimal strategy for the parents would be to adopt a cooperative attitude leading to a win-win solution [Folberg et al. 2005, 79–80]. The process of transformation of the dispute may be difficult in this context due to emotional distress and mutual lack of trust of the parties. This opens a rational possibility of adoption of either a classical conflict resolution procedure (for instance, mediation), or, if simultaneous presence of parties is either not possible or not advisable, a usage of negotiation support systems as for instance the PPSS may be the justified choice. The system described in the present contribution aims at facilitation of communication between the parties, providing information about the scope of decisions that have to be made to finalize the work on the agreement, establishing relations between parents and their children and, finally, at contribution to the decision-making process of the parties.
3.
The Structure of Parent Plans ^
4.
Knowledge Representation in the PPSS ^
The PPSS knowledge system is based on propositional elements that are represented by both CBR structures (dimensions and factors) as well as by defeasible rules. Almost all of this knowledge is representable in predicate logic which is compatible with the possibility of PROLOG implementations. This Section is devoted to presentation of each type of knowledge representation structure in the system (cf. [Araszkiewicz, Łopatkiewicz, Zienkiewicz 2013a and 2013b]) with particular focus on CBR structures and their discussion in the light of relevant AI and Law literature.
∀o ∃Qi (o ∊ Qi), and
∀oi ∀Qi, Qj (((oi ∊ Qi) ∧ (oi ∊ Qj)) → (Qi = Qj)).
A Parent Plan (PP) is a subset of O (PP ⊂ O), which is a selection of single elements from Questions:
∀oi ((oi ∊ PP) iff ∀Qi ((oi ∊ Qi) → (¬ ∃oj ((oj ∊ Qi) ∧ (oi ≠ oj))))
A Parent Plan is complete if and only if the following condition holds:
∀Qi, Qi ⊂ O, ∃ oi ((oi ∊ PP)).
Let Inc be a binary relation ranging over O (Inc ⊂ O2). We state that a Parent Plan is consistent if and only if the following condition holds:
∀oi, oj ((oi, oj ∊ PP) → (¬(Inc) oi, oj)).
A set of Options that is presented to the parent while constructing a concrete Parent Plan is referred to as a feasible set of Options.
Dimensions. The systems contain a library of dimensions: scalable knowledge representation structures [Ashley 1990]. Each dimension has two extreme points that represent states of affairs supporting opposite outcomes (classically, these extreme points may favor either a plaintiff or a defendant in the proceedings before the court; however, in the present system the extreme points of each Dimension point to the least and most favorable outcome as regards the realization of the well-being of the child). Each Dimension in the PPSS comprises the following elements:
- The name of the Dimension (identical to the name of Category C)
- A scale encompassing Valuations of the Dimension:
- Unsatisfactory (0)
- Sufficient (1–3)
- Good (4–6)
- Excellent (7–9).
- The alternative sets of Options that by default qualify as realization of the value of the child’s well-being.
There are 10 Dimensions in the system, one for each basic Category of the system. There are important differences regarding this account of Dimensions and the classical account developed by Rissland and Ashley, most authoritatively presented in [Ashley 1990]. The first important difference is that the extreme points of the dimensions do not support the opposing parties. The second difference concerns the structure of the dimension, which is much more complex here. Each point on the scale of the dimension may be satisfied by different Options (or sets of Options). Indeed, there may be two or more different Options that will be classified on a given point of scale of a given Dimension. The third difference is that each Parent Plan must be indexed by all of the Dimensions (in HYPO for instance, a case could be characterized by one or more dimensions). Fourth, the function of Dimensions here is not to give a basis for any analogical reasoning, but rather to represent the degree to which the value of the child’s well-being is satisfied by a given Parent Plan.
Defeasible Rules Set. The system comprises a huge (>100) set of defeasible rules extracted from the knowledge on pedagogy and developmental psychology as well as from the corpus of judgments of the PSC related to the concept of the well-being of the child. These rules are functions which accept Options as input and valuation on the scales given by Dimensions as output. As Options may be incompatible with each other, the system consists also of the rules for priority assignment between conflicting defeasible rules [Prakken and Sartor 1998].
Environmental Factors Set. The set of Environmental Factors (EFs) encompasses contextual information that is introduced by the users of the system in the initial stage of its operation. The factors employed here are binary knowledge representation structures: they may be either present or absent in a given case [Aleven 1997]. This set encompasses data such as distance from the residence of each parent to the schools in the surrounding area, the parents’ monthly salaries, as well as relevant data concerning the parents’ and the child’s personal characteristics. Like Options, there are Environmental Factors that are mutually incompatible (as for an obvious example, one child cannot have different ages at the same time). Certain EFs may also be incompatible with certain Options. Importantly, EFs may change the valuation function expressed by the default Defeasible Rules.
5.
Using the System. A Study of An Example ^
Example 1.
O (1). No contact with grandparents.
O (2). Rare contact with grandparents – a few times a year (for instance during holidays, family meetings).
O (3). More frequent contact (up to a few times a month).
O (4). Frequent contact (up to a few times a week), both planned and spontaneous.
O (5). Daily contact (including temporal exercise of custody by grandparents while parents are absent).
█
Example 1 continued.
R1: O (1) => U (0)
R2: O (2) => S (1)
R3: O (3) => G (4)
R4: O (4) => G (6)
R5: O (5) => E (9)
This assignment may seem arbitrary, and to a certain extent this opinion would be justified. Such situations almost always occur when measure scales are applied to legal contexts [Alexy 2003]. However, in this context, the valuation is suggested by the exact wording of the judgment of the PSC, III CZP 42/88, which is quoted further in later investigations.
█
Example 1 continued.
- I AGREE WITH THIS ASSESSMENT
- I DISAGREE WITH THIS ASSESSMENT
- THE GRANDPARENTS DON’T KNOW THE CHILD
- THE GRANDPARENTS ARE NOT WILLING TO HAVE CONTACT WITH THE CHILD
- THE GRANDPARENTS COULD HAVE A BAD INFLUENCE ON THE CHILD OR TYPE IN YOUR OWN REASON.
6.
Conclusion ^
7.
References ^
Aleven, Vincent, Teaching case-based argumentation through a model and examples. (Unpublished doctoral dissertation). University of Pittsburgh Graduate Program in Intelligent Systems (1997).
Alexy, Robert, On Balancing and Subsumption. A Structural Comparison, Ratio Juris Vol. 16, pp. 433–449 (2003).
Ashley, Kevin, Modeling legal argument: Reasoning with cases and hypotheticals. MIT Press, Cambridge, Mass (1990).
Araszkiewicz, Michał/Łopatkiewicz, Agata/Zienkiewicz, Adam, Factor-Based Parent Plan Support System. In: Proceedings of the 14th International Conference on Artificial Intelligence and Law (ICAIL 2013), (Francesconi, E., Verheij B. eds.), ACM, New York, pp. 171–175 (2013a).
Araszkiewicz, Michał/Łopatkiewicz, Agata/Zienkiewicz, Adam, Parent Plan Support System – Context, Functions and Knowledge Base. In: Abramowicz W. (ed.), Business Information Systems Workshops, Lecture Notes in Business Information Processing Vol. 160, Springer, pp. 160–171 (2013b).
Folberg, Jay/Golann, Dwight/Stipanowich, Thomas/Kloppenberg, Lisa, Resolving disputes: Theory, practice, and law. Aspen Publishers, New York (2005).
Lodder, Arno/Zeleznikow, John, Developing an online dispute resolution environment: Dialogue tools and negotiation support systems in a three-step model. Harvard Negotiation Law Review Vol. 10, pp. 287–337 (2005).
Lynch, Colin/Ashley, Kevin/Pinkwart, Neil/Aleven, Vincent, Concepts, structures and goals: Redefining ill-definedness. International Journal of Artificial Intelligence in Education Vol. 19, pp. 253–266 (2009).
Prakken, Henry/Sartor, Giovanni, Modelling Reasoning with Precedents in a Formal Dialogue Game. Artificial Intelligence and Law Vol. 6, pp. 231–287 (1998).
Zeleznikow, John/Stranieri, Andrew/Gawler, Mark, Project report: Split-Up: A legal expert system which determines property division after divorce. Artificial Intelligence and Law Vol. 3, pp. 267–275 (1996).
Michał Araszkiewicz
Adjunct, Jagiellonian University, Faculty of Law and Administration, Department of Legal Theory
Bracka 12, 31-007 Kraków, PL
Agata Łopatkiewicz
PhD researcher, Faculty of Philosophy, Institute of Education
Batorego 12, 31-135 Kraków, PL
Adam Zienkiewicz
Adjunct, University of Warmia and Mazury, Department of Theory and Philosophy of Law and State
Warszawska 98, 10-702 Olsztyn, PL