BuST-Bundled Suffix Trees

Abstract

We introduce a data structure, the Bundled Suffix Tree (BuST), that is a generalization of a Suffix Tree (ST). To build a BuST we use an alphabet Σ together with a non-transitive relation ≈ among its letters. Following the path of a substring β within a BuST, constructed over a text α of length n, not only the positions of the exact occurrences of β in α are found (as in a ST), but also the positions of all the substrings β₁ , β ₂ , β₃ , . . . that are related with β via the relation ≈ among the characters of Σ , for example strings at a certain ”distance” from β . A BuST contains O(n^1+δ ) additional nodes (δ < 1) in probability, and is constructed in O(n^1+δ ) steps. In the worst case it contains O(n²) nodes.