A New Parallel Partition Algorithm for Parallel Suffix Tree Construction

The suffix tree is a compacted trie of all suffixes of a given string. It is a fundamental data structure in a wide range of domains such as text processing, data compression, computer vision, computational biology, and so on [1]. Moreover, it can be used for network researches such as web analysis, which has been studied actively [2], [3]. For example, suffix trees have been utilized to effectively search for genomic DNA data in databases or text in the web. Recently, as parallel architectures such as distributed systems and CMPs improve, there have been studied developing parallel suffix tree construction algorithms practically. Chen and Schmidt [4] proposed a parallel algorithm for constructing suffix trees on a computational grid. Tsirogiannis and Kouds [5] proposed cacheconscious suffix tree construction algorithms that are tailored to CMP architectures. These algorithms take an approach of dividing suffixes into a number of partitions and constructing parallel the suffix tree for each partition. In this paper, we propose a new algorithm of partitioning suffixes. Basically, our algorithm is similar to that in [5], which partitions suffixes by prefixes of variable lengths. However, we use a trie instead of a hash table as an auxiliary data structure. Moreover, our algorithm can be easily parallelized to adapt to parallel architectures.