Fast Average-Case Pattern Matching on Weighted Sequences

TL;DR

This paper introduces algorithms for weighted string matching that achieve average-case sublinear search times, significantly improving efficiency for uncertain sequences with specific weight ratios.

Contribution

It presents novel algorithms for weighted pattern matching with average-case sublinear search times under certain weight ratio conditions.

Findings

Algorithms achieve average-case search time o(n) and o(σn)

Work in linear preprocessing time and space

Applicable for specific weight ratio thresholds

Abstract

A weighted string over an alphabet of size $\sigma$ is a string in which a set of letters may occur at each position with respective occurrence probabilities. Weighted strings, also known as position weight matrices or uncertain sequences, naturally arise in many contexts. In this article, we study the problem of weighted string matching with a special focus on average-case analysis. Given a weighted pattern string $x$ of length $m$, a text string $y$ of length $n>m$, and a cumulative weight threshold $1/z$, defined as the minimal probability of occurrence of factors in a weighted string, we present an algorithm requiring average-case search time $o(n)$ for pattern matching for weight ratio $\frac{z}{m} < \min\{\frac{1}{\log z},\frac{\log \sigma}{\log z (\log m + \log \log \sigma)}\}$. For a pattern string $x$ of length $m$, a weighted text string $y$ of length $n>m$, and a cumulative weight threshold $1/z$, we present an algorithm requiring average-case search time $o(\sigma n)$ for the same weight ratio. The importance of these results lies on the fact that these algorithms work in average-case sublinear search time in the size of the text, and in linear preprocessing time and space in the size of the pattern, for these ratios.