A family of fast exact pattern matching algorithms

TL;DR

This paper introduces a family of comparison-based exact pattern matching algorithms that leverage multi-dimensional arrays to process multiple text windows simultaneously, improving average time complexity especially for short patterns and medium-sized alphabets.

Contribution

The paper presents novel pattern matching algorithms that outperform existing methods like Boyer-Moore-Horspool in specific parameter ranges, particularly for natural language text search.

Findings

Algorithms outperform Boyer-Moore-Horspool in many cases

Best performance for alphabet size 30-40 and pattern length 4-10

Effective for search in natural language text databases

Abstract

A family of comparison-based exact pattern matching algorithms is described. They utilize multi-dimensional arrays in order to process more than one adjacent text window in each iteration of the search cycle. This approach leads to a lower average time complexity by the cost of space. The algorithms of this family perform well for short patterns and middle size alphabets. In such case the shift of the window by several pattern lengths at once is quite probable, which is the main factor of algorithm success. Our algorithms outperform the Boyer-Moore-Horspool algorithm, either in the original version or with Sunday's Quick search modification, in a wide area of pattern length - alphabet size plane. In some subareas the proposed algorithms are the fastest among all known exact pattern matching algorithms. Namely, they perform best when alphabet size is about 30-40 and pattern length is about 4-10. Such parameters are typical for search in natural language text databases.