Information Technology Reference
In-Depth Information
methodologies for problem solving and/ormethods for knowledge representing and
reasoning, in particular to think up the legal procedures. In general, high-level agents
monitor and set the behaviour ofthe low level ones. Low-level agents act on the ob-
ject-level data and knowledge, that make the extensions of the functions that describe
the universe of discourse.
Let us take as example anegotiation process:high-level agents guide the process
and determine when a new round should start, or finish; low-level ones have the
autonomy to choose the actions to be performed, according to the state of the domain
of discourse. That is, depending on the domain, low level agents will compute a valid
output in each round.All the agents and their roles are depicted in [11]. These agents
were defined according to the specific requirements of the project, following an itera-
tive cut-down practiceof increasing specification.
3.2 The Ontology
Having specified the high level architecture, let us now show how it can be used to
implement specific services. This will be done through the use of ontologies. The
method consists in defining a specific ontology for each legal domain. Each ontology
encodes a domain theory, actions, constraints and rules, revealing to the agent what
action is to be executed and how to achieve it, i.e., according to the legal domain
being attended to. The main advantage of this approach is that a single agent can be
used to perform a similar task in a wide range of domains in opposition to a tradi-
tional one in which an agent would be used for each different domain. As an example,
let us consider the action of searching for similar cases. Instead of having three differ-
ent agents, one for each of the legal domains under equation (i.e., Commercial, Fam-
ily and Labour), we have one single agent and three different ontologies.This line of
attack significantly increases functionality reuse and allows for a single architecture
supporting services in a wide range of domains. It also simplifies the task of extend-
ing the architecture for addressing new legal domains by developingan ontology for-
everyone, with all the actions, rules, constraints and specific theories. This will tell the
low-level agents how to act in a new domain.
Each ontology comprises four components: the vocabulary, the actions, the fea-
tures and a theory. The vocabulary contains all the words that can be used to describe
the entities that belong to the domain. Actions define how each action should be exe-
cuted according to the domain of the ontology. The theory will define all the elements
that make up the perceptions about the universe of discourse (e.g., type, number).
Finally, features allow for add up of invariants to the perceptions and actions that
make up the ontology.
As an example, let us consider the computation of the BATNA and WATNA in
two different legal domains, namely: Commercial Law and Labour Law. These
twoabstract conceptions denote, as it was described above, the best and worst possible
scenario in a litigation process, i.e., it has a meaning that can be of use independently
of the domain of the dispute. Nevertheless, depending on that domain, it is computed
in different ways. In Labour Law we will have to consider views in terms of worker
antiquity, monthly salary, seniority, a joust cause for dismissal, just to name a few. On
the other hand, in Commercial Law key views would be the date in which the product
Search WWH ::
Custom Search