Automated reasoning techniques are crucial for situations in which a huge number of heterogeneous reasoning rules must be taken into account. They allow both ensuring the coherence of the system and making the decision process equitable and more efficient. The National Bureau for Family Allowance of the Grand-Duchy of Luxembourg is responsible for the attribution of allowances to more than 160,000 individuals whose cases, due to the peculiarity of the local economy based on foreign laborers, and given the European and bilateral agreement between countries, pertain to different legislations. This paper presents Cadral, a decision support system under development for processing the allowance applications. The system mixes an inference engine based on the Soar forward-chaining architecture with an interpreter for easy-to-write-behavior rules, so that a non-computer specialized user can update the system, according to the evolution of the law. The rules record administrative procedures used for the processing of the applications, while links to a legal database, used in connection with the reasoning trace of the system, allows exhibiting a legal justification of the resulting decisions.
[1]
M. M. Groothuis,et al.
Expert system support and juridical quality
,
2000
.
[2]
Eric McCreath,et al.
Building and Testing the SHYSTER-MYCIN Hybrid Legal Expert System
,
2003
.
[3]
Clare Bates Congdon,et al.
The Soar User''''s Manual
,
1986
.
[4]
Allen Newell,et al.
SOAR: An Architecture for General Intelligence
,
1987,
Artif. Intell..
[5]
Michael R. Fehling.
Unified Theories of Cognition: Modeling Cognitive Competence
,
1993,
Artif. Intell..
[6]
T. Gordon.
From Jhering to Alexy { Using Arti cial Intelligence Models in Jurisprudence
,
1994
.
[7]
Charles L. Forgy,et al.
Rete: a fast algorithm for the many pattern/many object pattern match problem
,
1991
.