Suffix trees are versatile data structures that are used for solving many string-matching problems. One of the main arguments against widespread usage of the structure is its space requirement. This paper describes a new structure called suffix vector, which is not only better in terms of storage space but also simpler than the most efficient suffix tree representation known to date. Alternatives of storage representations are discussed and a linear-time construction algorithm is also proposed in this paper. Space requirement of the suffix vector structure is compared to the space requirement of alternative suffix tree representations. We also make a theoretical comparison on the number of operations required to run algorithms on the suffix vector.
Krisztián Monostori, Arkady B. Zaslavsky, H