Performance assessment of dead-zone single keyword pattern matching

Abstract

A so-called dead-zone pattern matching family of algorithms has previously been proposed as a concept. Here the performance of several instances of the family are empirically investigated. An abstract description of the algorithm family is given, as well as of these instances. This leads to a total of five different implementations of the algorithm. The time-performance of each of these implemented variants was then compared against those of the well-known Horspool and Boyer-Moore pattern matching algorithms, the latter being representative of the best-known pattern matchers currently available. Various data sets, pattern selections and pattern lengths were used, but the same Horspool-like shift tables were used throughout by the dead-zone variants. It was found that in some contexts variants of the dead zone algorithm compare very favourably against the Horspool and Boyer-Moore algorithms.