We have investigated the problem of clustering documents according to their semantics, given incomplete and incoherent hints reflecting the documents’ affinities. The problem has been rigorously defined using graph theory ...