Multi-objective optimization problems with more than three objectives, which are also termed as many objective optimization problems , play an important role in the decision making process. For such problems, it is computationally expensive or even intractable to approximate the entire set of optimal solutions. An alternative is to compute a subset of optimal solutions based on the preferences of the decision maker. Commonly, interactive methods from the literature consider the user preferences at every iteration by means of weight vectors or reference points. Besides the fact that mathematical programming techniques only produce one solution at each iteration, they generally require first or second derivative information, that limits its applicability to certain problems. The approach proposed in this paper allows to steer the search into any direction in the objective space for optimization problems of discrete nature. This provides a more intuitive way to set the preferences, which represents a useful tool to explore the regions of interest of the decision maker. Numerical results on multi-objective multi-dimensional knapsack problem instances show the interest of the proposed approach.