Evaluation of recursive queries and computing transitive closures require multiple accesses to the involved relations. In a federated database this leads to multiple accesses to the participants of the federation. Since the components are not uniform in terms of computation power, reliability, and communication delays, it might be desirable to minimize the number of accesses to the individual databases, and to maximize the size of the obtained answer with respect to time. Based on this observation, we developed cooperative query planning methods, termed Deep Federated Semi-Naive (DFSN), for computing the strong partial transitive closure of a relation. We have implemented and tested these algorithms in a real database environment. The experimental results show better performance of the DFSN methods over the conservative semi-naive approaches in that they produce large answer sets in time that is considerately shorter than the time needed by the conservative approaches.