I am solving a combinatorial optimization problem, where I do not have a global optimum, so the goal is to improve the objective function as much as possible. So, to do this, I was inspired by this article Reactive Search strategies using Reinforcement Learning, local search algorithms and Variable Neighborhood Search, I apply during several iterations, heuristics to improve the solution, that is to say, that at each iteration I must choose a heuristic and apply it on the current solution.

In this article, they have defined the state space as the set of heuristics to apply and the action space is the choice of a heuristic among these heuristics.

Regarding the reward, they gave +1 if the solution is improved and -1 if the solution is not improved.

Sincerely, I did not understand how we define the reward for example here -1 and 1, and according to which criteria we choose the reward to use?