Supporting Evacuation Missions with Ontology-Based SPARQL Federation

We study ontology-based SPARQL federation in support of coordinated action by deployed units in military operations. It is presumed that bandwidth is limited and unstable. Thus, we need an approach that generates few HTTP requests. Existing techniques employ join-order heuristics that may cause requests to multiply as a factor of the number of joins in a query. This can easily lead to an amount of traffic that exceeds network capacity. We propose an approach that builds an inmemory excerpt of the remote sources, sending one request to each source. A query is answered against this excerpt, which is a provably sound and complete representation of the sources wrt. query answering. The paper ends with a case study involving three military sources used for planning evacuation missions.