Longest-Common-Prefix Computation in Burrows-Wheeler Transformed Text (2006)
BibTeX
@MISC{Umar06longest-common-prefixcomputation,
author = {Ibrahim Umar and Rosni Abdullah},
title = {Longest-Common-Prefix Computation in Burrows-Wheeler Transformed Text},
year = {2006}
}
Bookmark
 |
 |
 |
 |
 |
 |
OpenURL
Abstract
In this paper we consider the existing algorithm for computation of the Longest-Common-Prefix (LCP) array given a text string and its suffix array and adapt it to work on Burrows-Wheeler Transform (BWT) text. We did this by a combination of pre-processing steps and improvement based on existing algorithm. Three LCP array computation algorithms were proposed, namely LCPB-A, LCPB-B and LCPB-C that need only BWT text as an input. LCPB-A was a simple adaptation from the existing algorithm which did a pre-processing on BWT text of length n to generate suffix array and its original text. Then, output of this step was fed to existing algorithm which does LCP array computation in O(n) time. LCPB-B reduces LCPB-A pre-processing time while requiring additional 4n space as it substitutes original text with two auxiliary arrays in LCP array computation. LCPB-C takes a step forward from LCPB-A and LCPB-B by utilizing the advantage of BWT text structure. It effectively reduces pre-processing time and memory requirements for a price of θ(δn) time when calculating LCP array, where δ is the average LCP. Experimental results showed that in terms of speed, LCP-A practically performs as well as the original algorithm, compared to LCPB-B and LCPB-C. However, LCPB-C consumes less space than the other algorithms.
