On the correctness of query results in XML P2P databases

In XML peer-to-peer (p2p) database systems, query results are usually assumed to be incomplete. Incompleteness issues derive from the unstable and open-ended nature of the network, where new nodes may connect at any time, and existing nodes may suddenly disappear. The incompleteness of input data used for query evaluation may also lead to the incorrectness of query results, which greatly affects the usefulness of the whole p2p approach to XML databases. We formally deal with the problem of result correctness in the presence of incomplete input data, and identify query classes for which the result correctness can be statically predicted at no extra cost w.r.t. usual syntactical and semantic query analysis.