This research proposes a methodology to solve distributed optimization problems where quasi-autonomous decision entities directly interact with each other for partial information sharing. In the distributed system we study the quasi-autonomy arising from the assumption that each decision entity has complete and unique responsibility for a subset of decision variables. However, when solving a decision problem locally, consideration is given to how the local decisions affect overall system performance such that close-to-optimal solutions are obtained among all participating decision entities. Partial information sharing refers to the fact that no entity has the complete information access needed to solve the optimization problem globally. This condition hinders the direct application of traditional optimization solution methods. In this research, it is further assumed that direct interaction among the decision entities is allowed. This compensates for the lack of complete information access with the interactive exchange of non-private information. The methodology is tested in different application contexts: manufacturing capacity allocation, single machine scheduling, and jobshop scheduling. The experimental results show that the proposed method generates close-to optimal solutions in the tested problem settings.