Abstract

In this work, a frame is posed which allows to define global search heuristics in an efficient and declarative way, which interacts with a specific computational implementation of a problem. An overview of different ways to solve problems by the use of global search is presented, followed by the specification of the language proposed. A real decision support system was developed through the use of the language. The problem faced was an extension of the capacity vehicle routing problem. The followed approach minimizes the development cost of a decision support system for logistic and productive environments, since the performance of different heuristics can be tested using the language in a straightforward way. Besides, when new requirements or additional knowledge about the problem appear, the solving engine can be easily modified through the heuristic language. © 1998 Elsevier Science Ltd All rights reserved.