We propose a 0/1 integer programming model to tackle the office space allocation (OSA) problem which refers to assigning room space to a set of entities (people, machines, roles, etc.), with the goal of optimising the space utilisation while satisfying a set of additional requirements. In the proposed approach, these requirements can be modelled as constraints (hard constraints) or as objectives (soft constraints). Then, we conduct some experiments on benchmark instances and observe that setting certain constraints as hard (actual constraints) or soft (objectives) has a significant impact on the computational difficulty on this combinatorial optimisation problem.