Using Case Based Reasoning and Principled
Negotiation to provide Decision Support for
Dispute Resolution
Davide Carneiro
a
, Paulo Novais
a
, Francisco Andrade
b
,
John Zeleznikow
c
and José Neves
a
a
Department of Informatics, University of Minho, Braga, Portugal
b
Law School, University of Minho, Braga, Portugal
c
School of Management and Information Systems, Victoria University, Melbourne, Australia
Abstract. The growing use of Information Technology in the commercial arena
leads to an urgent need to find alternatives to traditional dispute resolution. New
tools from fields such as Artificial Intelligence should be considered in the process
of developing novel Online Dispute Resolution platforms, in order to make the
ligation process simpler, faster and conform with the new virtual environments. In
this work, we describe UMCourt, a project built around two sub-fields of Artificial
Intelligence research: Multi-agent Systems and Case-based Reasoning, aimed at
fostering the development of tools for Online Dispute Resolution. This is then used
to accomplish several objectives, from suggesting solutions to new disputes based
on the observation of past similar disputes, to the improvement of the negotiation
and mediation processes that may follow. The main objective of this work is to
develop autonomous tools that can increase the effectiveness of the dispute
resolution processes, namely by increasing the amount of meaningful information
that is available for the parties.
1. Introduction
In the words of Abraham Lincoln: “Discourage litigation, persuade your neighbour to
compromise where you can. Point out to them how the nominal winner is often the loser... in
expenses and waste of time”. This is one of the sentences that better describes the need for an
alternative to traditional courts and the appearing of Alternative Dispute Resolution (ADR) [1].
ADR includes mechanisms that aim to solve disputes without recurring to the traditional judicial
process, i.e. courts. ADR methods are nowadays broadly used by both the legal system and the
parties involved as the first step in the path towards the dispute resolution. There are even
countries in which parties are encouraged or required to try an alternative way of dispute
resolution before advancing into a court.
One of the factors that definitively boosted the acceptance of ADR is the current workload of
courts. Often it takes many years for a case to receive a final hearing. Thus, a new approach is
required to diminish the number of cases that actually need to be resolved by traditional court
processes. Parties, in particular, also look at ADR tools in search for lower expenses, more
privacy and less exposure than the provided by traditional trials. Another advantage is that the
entity that will decide a given case can be agreed on by the parties instead of being mandatorily
assigned, which in a certain way increases the confidence and overall satisfaction of the
disputing parties in the result obtained.
An alternative way of solving disputes arising out of the contractual performance in virtual
environments is that of Online Dispute Resolution. This new paradigm allows the moving of the
traditional alternative dispute resolution methods from a physical to virtual place [24]. Hence,
parties not only reduce litigation but gain a simple and efficient way of dealing with disputes,
saving both temporal and monetary costs [33]. Several methods of ODR may be considered,
from negotiation and mediation to variations of arbitration or modified jury proceedings [34].
Indeed, the growing use of Information and Communication Technologies (ICT) in the
commercial arena lead to an urgent need to find alternatives to dispute resolution. The
traditional paper-based courts and arbitration, designed for the industrial era, are now outdated.
In fact, Civil and Commercial Law are still turned towards a type of society and a way of doing
business that is related to a mostly industrial society and the traditional use of the paper as the
primary support tool. Given the rising use of Information Technology, electronic means for
resolving disputes need to be considered. This should include emerging tools from fields such as
Artificial Intelligence that are used for effective idea and strategy generation. This will result in
a more creative process, rich in ideas and solutions, that will ultimately lead to simpler, faster
and more effective dispute resolution processes.
In this sense, the use of software agents that embody intelligent techniques can become
particularly interesting considering their ability to recognise and evaluate facts, positions and
relevant information. Moreover, the use of these agents, as building blocks of distributed
platforms that support ODR processes, has to be considered. Of major importance in this
context are systems that are able to simulate and calculate outcomes of disputes, establish
negotiation paths to achieve the desired objectives and warn the parties about the possible
alternatives to an agreement.
There is a variety of ODR systems, including legal knowledge based systems and systems
that help settle disputes in an online environment [35]. Second generation ODR, in which ODR
systems act as an autonomous agent [36] are a most appealing way of solving disputes. Also
interesting here is the view of Katsh and Rifkin concerning the role of technological tools. The
authors see technology that works with the mediator or arbitrator as the fourth party in the
dispute, together with the two disputing parties and the third neutral [37]. In fact, there has been
a recent tendency to foster the intervention of software agents, acting either as decision support
systems [24] or as real electronic mediators [36]. Surely, this latest role for software agents
could benefit from the use of Artificial Intelligence techniques such as Case-based Reasoning
(CBR) and information and knowledge representation models. Models of the description of the
fact situations, of the factors relevant for their legal effects allow the agents to be supplied with
both the static knowledge of the facts and the dynamic sequence of events [36].
Evidently, representing facts and events is not sufficient for a dispute resolution. In order for
the software agent to perform actions of utility for the resolution of the dispute, it also needs to
know the terms of the dispute and the rights or wrongs of the parties [36]. Moreover, it is of
utter importance that the software agent is able to foresee the legal consequences of the said
facts and events. We thus need to consider whether agents can evaluate the position of the
parties and present them with useful proposals, taking into a consideration of which of the two
parties would have a higher probability of being penalised or supported by a judicial decision of
the dispute and, therefore, who would be more or less willing to make concessions in their
claims [16]. The ability to understand the position of the parties is vital for the successful
involvement of software agents in the process. To do so, it is mandatory for the software agent
to have the characteristics of consistency, transparency, efficiency and enhanced support for
dispute resolution, in order to allow it to replicate the manner in which decisions are made and
thus make the parties aware of the likely outcome of litigation [3]. That is to say, software agent
intervention in ODR procedures should take into account the alternatives, for the parties, to an
ODR negotiated agreement.
Thus, in such a context, it would be interesting to consider some useful and well known ADR
concepts such as the BATNA, or Best Alternative to a Negotiated Agreement. This concept
denotes the best scenario possible if the negotiation process fails, i.e., if the case was to go into
a court, what would be the best outcome? Knowing their BATNA helps parties to achieve an
outcome using negotiation that is hopefully better than it would be if achieved through litigation
[45]. The importance of this concept is well expressed by Goldberg et al. in [31]. In fact, if a
party is unaware of what results could be obtained if the negotiations are unsuccessful, he runs
the risk of: entering into an agreement that he would be better off rejecting; or rejecting an
agreement he would be better off entering into.
Following the same line of thought the WATNA, which denotes the worst possible outcome
along a litigation path, should also be considered [32]. It can also be quite relevant in
complementing principled negotiation [46] with a justice or rights based approach, and thus
leading to a calculation of the real risks that parties will face in judicially determined litigation,
imagining the worst possible outcome. In the same sense, considering the MLATNA – Most
Likely Alternative to a Negotiated Agreement, may also be useful as a way to show parties the
alternative that is most likely to happen if the negotiated process fails. These three concepts are
definitively of utter importance for the parties to have a clear general picture of what the
alternatives are, thus supporting wiser decisions. Finally, one last important concept is the one
of ZOPA – Zone of Potential Agreement, which denotes the range of the possible outcomes,
i.e., what may happen at the end of the process, and is delimited by the WATNA and BATNA
[4].
Considering the usefulness of these concepts in ODR, it becomes obvious that these systems
will go much further than just transposing ADR ideas into ODR environments. They should
actually be guided by judicial reasoning, getting disputants to arrive at outcomes in line with
those a judge would reach [14]. Despite there being difficulties to overcome at this level, the
generalised use of software agents as Decision Support Systems points out the usefulness of
following this path.
Taking all of this into consideration, the work described in this paper focuses on two key
points. First, the use of Artificial Intelligence techniques that enable not only contextualized
information retrieval but also allow the system to evolve according to the results of this
retrieval. Particularly, we will be analyzing the role that Case-based Reasoning can play not
only in the information retrieval task but also on improving other processes. Second, we will be
looking at the use of traditional techniques, specifically negotiation and mediation, as a way to
improve the efficiency of dispute resolution. These two trends merge to fulfill a unique
objective: to provide contextualized information and guidance processes to disputant parties so
that they can solve their dispute out of courts.
1.1 Related Work
Proving the validity and relevancy of this research field, a group of interesting projects with the
objective of giving birth to dispute resolution mechanisms can be pointed out. In this section we
describe some of the projects which intersect with this work.
Rule-based Legal Decision-making System (LDS) dates from 1980 and was one of the first
negotiation support systems to be developed [50]. The domain of this system was liability law.
This field of law holds responsible product distributors and manufacturers for the injuries their
products may cause. The system created embodied the skills and knowledge of a human expert
in the shape of antecedent-consequent rules. The project had the objective of formalizing the
decision-making processes of attorneys and claims adjusters involved in product liability
litigation in the shape of rule-based models so that the effects that changes in legal doctrine have
in settlement strategies and practices could be studied. The authors formalized the strict-liability
concept on ROSIE language, so that the defendant could or could not be considered liable.
A more recent project, focusing on providing support for decisions is EXPERTIUS. This is a
decision-support system that advices Mexican novice judges and clerks upon the determination
of whether the plaintiff is or not eligible for granting him/her a financial pension (on the basis of
the “feeding obligation”) and if so upon the determination of the amount of that pension [51].
The system has three main modules: a tutorial module, an inferential module and a financial
module. The tutorial module guides the user through the accomplishment of several tasks. The
inferential module evaluates evidence based on weights that the user assigns to each piece of
evidence. It determines which prepositions are defeated and which prevail. At last, the financial
module assists the user on the calculus of the value of pensions according to some criteria.
For reasoning in these terms, EXPERTIUS has an extensive way of representing the
knowledge about the several parameters. Judicial expert knowledge was represented as having
three interrelated levels: one for representing the expert knowledge, one for representing the
decisions internal to each procedural stage as regulated by procedural law and a third one that
corresponds to the dialogical confrontation pattern of the case that arises simultaneously to the
decisions taken in the intermediate level.
The knowledge embodied in the system was divided into five layers. The formal doctrine
contained rules from the legislation and common law. The informal principles contained rules
that are not explicitly expressed in the law but are generally agreed upon by legal practitioners.
Under the strategies layer the authors coded the methods used by legal practitioners to
accomplish a given goal. The subjective considerations layer contains rules that anticipate the
subjective responses of people involved in legal interactions. At last, the secondary effects layer
contained rules that describe the interactions between rules. The authors concluded that despite
the number of rules needed for formalizing the law and the strategies, the rule-based model was
feasible and suited for this particular domain.
In the field of negotiation support, a major work is the Family_Winner project, which
integrates game theory and heuristics in the field of family law, more specifically, in disputes
arising from divorce processes [2]. The research is largely based on the Adjusted Winner
algorithm [2], an algorithm that merges insights from the fields of game and decision theory.
This algorithm aims at the distribution of a group of items or issues by the parties, according to
a mechanism of point allocation. In this process, each party quantifies how much he/she wants
the item and whoever values more a given item receives it.
Still in the negotiation support, another project worth mentioning is SmartSettle [48]. It is an
online negotiation system that can be described as a generic tool for decision-makers. It is
intended for parties with conflicting objectives that wish to reach a formal agreement. For
achieving it, tools that help to define the interests and possible trade-offs are used, with an
emphasis on recognizing the parties’ satisfaction in the search for optimal global solutions. This
platform can be used to solve problems relating to family, insurance, real estate, labour-
management, contract negotiations, among others.
When comparing the work presented in this paper with the related work mentioned in this
section, the main difference that can be noted is that, contrary to the most common approach,
we do not aim to provide support for dispute resolution in court. We rather aim for providing
valuable information for the parties and eventually the neutrals, so that better and more
informed decisions can be taken. While doing so, we also expect disputing parties from going
into court, by making them aware of the possible consequences, fostering cooperative off-court
processes. Nevertheless, the tool presented can still be used throughout a complete dispute
resolution process, ranging from the initial providing and compilation of information, to the
definition of an outcome. Another major difference is that in this work, there is a close
integration between the phase of generating information and the phase of actual dispute
resolution through traditional means. That is, the information that is compiled to inform the
disputing parties is also used to support and help define the mediation or negotiation process
that may follow.
2. Agents and Negotiation
As mentioned before, there is a particular interest in implementing ODR tools that behave the
same way a human would. Particularly interesting is the use of software agents to implement
social interaction processes, specifically negotiation. Generally, negotiation is classified as
distributive or integrative [5].
In distributive negotiation, one looks at the problem as something that can be divided and
distributed by the parties in an attempt to maximize their satisfaction. An example scenario is
the classical winding up of a company in which the assets of the company are sold and the
proceedings collected are used to discharge the liabilities. In game theory this is known as a
zero-sum game. Two important concepts here are the ones of utility and resistance [6]. Utility
denotes the value that a given item has to a party while resistance denotes the willingness of a
party to change the utility of an item. A good negotiator usually tries to convince the other party
that certain items do not have the value that they are given. The negotiator will succeed if the
opponent has a low resistance in that item and if he does so, it will be easier for the negotiator to
win that item or he will, at least, be in a better position for the rest of the negotiation process [7].
Accordingly, one can define utility functions that help to understand how each party values the
items being distributed, therefore predicting possible outcomes and the evolution of the
negotiation process [8].
In integrative negotiation, the problem is expected to have more solutions than the ones
visible at first sight. In these types of problems, the parties try to bring to the table as much
interests as possible so that there are more and more valuable items with which to negotiate.
When the parties are increasing the value of what they put in the table, they take into account
their interests which include the needs, fears, concerns, and desires. This type of negotiation is
also known as interest-based, as parties try to combine their interests and find common points in
which they are satisfied. By doing so, more satisfactory outcomes are achieved by both parties.
This makes integrative negotiation more desirable than distributive.
A good example for illustrating this difference is an old story about a brother and a sister both
wanting the last orange. To solve the dispute, their mother initially gives both children half the
orange. When she asks them their goals, she finds that the brother wants the orange for the juice
to drink while the sister wants the rind to flavor an orange cake. So the mother gives the sister
the rind and the brother the rest and they both get what they desire [49]. The outcome is
optimum. An important concept in this field is the one of Pareto efficiency [9]. In this case, the
solution obtained with the integrative approach would have been a Pareto efficient solution as
no better solution could have been achieved.
A key factor in dispute resolution is thus, as seen in the previous example, to identify the
interests of each party so that their positions can be better understood. Such a process can
eventually lead to what is known in game theory as a win-win game, i.e., all the parties are
better at the end of the negotiation process than when it started.
2.1. Principled Negotiation
One of the most effective methods for resolving conflicts is through negotiation. Specifically,
Principled Negotiation was developed by the Harvard Negotiation Project [46] and focuses on
the notion that parties look for mutual gains. When interests conflict, Principled Negotiation
advocates parties arrive at a ruling that is independent of the beliefs of either side. Principled
Negotiation puts forward five key points:
1) Separate the people from the problem. Personal matters or stronger personalities can often
influence the outcome of dispute resolution processes. This is especially significant in
scenarios in which the disputant parties have or had a personal relationship, such as in family
law disputes. Solving disputes by leaving aside personal affairs can avoid biasing a solution
towards a more influencing party.
2) Focus on interests, not on positions. It is important for a party to understand what their
interests are rather than blindly defending a position that may not be that solid. By isolating
the reasons why a position is most appealing, participants in a negotiation will increase the
chance of achieving agreement.
3) Invent options for mutual gain. Although parties have divergent objectives or interests, they
might still find positions that encompass mutual gain. Once interests have been ranked to
determine the relative importance of each, a range of options is discussed before deciding on
an outcome. Next, the negotiators need to invent options for mutual gain.
4) Insist on objective criteria. There are negotiation scenarios that cannot be treated as a win-
win situation. In such cases, unbiased independent evaluations should provide an outcome or
solution that both parties will agree on.
5) Know your Best Alternative to a Negotiated Agreement. When two parties negotiate they aim
at achieving a better result than would otherwise occur. Being unaware of what results one
could obtain if the negotiations are unsuccessful, one runs the risk of:
a. Entering into an agreement that would be better off rejecting; or
