Local Search (LS), is an iterative methodology to solve combinatorial optimization problems. The principle is to generate a first solution, and to iteratively perform slight modifications to this solution in order to obtain a good score with respect to the objective function.
Very Large-Scale Neighborhood (VLSN) is a sophisticated technique in Local Search to perform many modifications to the solutions at each iteration. These techniques have a greater visibility at each iteration and choose the next solution more efficiently. Very Large-Scale Neighborhoods have been successfully applied on many complex real-life problems.
VLSN are very hard to implement, but a recent PhD thesis has demonstrated a framework that expresses VLSN search algorithms in terms of high-level components [Mout11].
VLSN search algorithms expressed by mean of this framework exhibits several benefits compared to existing approaches : 1) a more natural design of VLSN search algorithm, 2) a better reusability of existing components, 3) a greater adaptability to modifications to the problem to solve, and 4) a faster development of complex VLSN approaches.