Abstract
The presented study deals with the so-called soft constraint satisfaction problem (SCSP) and proposes an extension to the standard SCSP formulation to accommodate a wider class of over-constrained situations and allow for a generally higher level of flexibility in the constraint-driven problem-solving. The extended modeling approach called Achievement-Weighted Constraint Satisfaction (AWCS) assumes the definition of constraint parameters ``traditional'' for SCSPs, as well as additional parameters specified to dynamically manipulate constraint weights in the course of solution search. These latter parameters make it possible to ``relax'' over-constrained models and obtain a solution even when there are mutually contradicting rules utilized by an AWCS problem-solver. To explore the proposed modeling framework, a task of finding an optimal route in car navigation, based on user preferences - a popular are of research in SCSP studies - is considered. A case study is presented, in which an optimal route is first modeled with constraints reflecting user preferences. Problem solutions having different optimality levels are then obtained. A software system is developed to automate both the optimal route modeling (via interaction with the user) and the solution search processes. The system is applied in an experiment conducted to validate the theoretical ideas. Experimental results are discussed, and conclusions are drawn.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have