Journal of Applied Sciences1812-56541812-5662Asian Network for Scientific Information10.3923/jas.2013.883.888AzizRafidah AbdulAyobMasriOthmanZalindaSarimHafiz Mohd62013136The basic idea of a Variable Neighborhood Search (VNS) algorithm is to systematically explore a neighborhood of a solution using a set of predefined neighborhood structures. Since different problem instances have different landscapes and complexities whilst different neighborhood structures may lead to different solution spaces, the choice of which neighborhood structure to be applied is a challenging task. Therefore, this work proposes an Adaptive Guided Variables Neighborhood Search (AG-VNS). AG-VNS has two phases. First, is a learning phase which is used to memorize neighborhood structures that can effectively solve specific soft constraint violations by applying the neighborhood structures to the best solution. These steps are repeated until a stopping condition for the learning phase is met. Second, is an improvement phase which is used to enhance the quality of a current best solution by selecting the most suitable neighborhood structure from memory that will be applied to the current solution. Its effectiveness is illustrated by solving course time tabling problems. The performance of the AG-VNS is tested over the Socha course time tabling datasets. Results demonstrated that the performance of the AG-VNS is comparable with the results of the other VNS variants, while outperforming some variants in particular instances. This demonstrates the effectiveness of applying the adaptive learning mechanism to guide the selection of the neighborhood structures in the VNS algorithm.]]>Thompson, J. and K. Dowsland,1995Abdullah, S.,2006Socha, K., M. Samples and M. Manfrin,2003Hansen, P. and N. Mladenovic,2001Abdullah, S. and H. Turabieh,2008Jat, S.N. and S. Yang,2009