Incremental nearest-neighbor browsing can avoid recomputing earlier results in spatial databases.

Distance browsing illustrates a broader advantage of incremental graph-style exploration. Hjaltason and Samet compare conventional k-nearest-neighbor retrieval with an incremental algorithm that returns the next neighbor without restarting the computation. Their experiments on R-trees found significant performance gains for distance-browsing queries, especially when another predicate limits acceptable results. The result does not prove that graph databases always outperform relational systems; it shows why adaptive traversal can suit queries whose depth or stopping point is unknown.