Efficient regular expression matching on LZ77 compressed strings using negative factors

Abstract

The state-of-the-art approaches for regular expression matching on LZ78 compressed strings do not perform efficiently. Moreover, LZ78 compression has some shortcomings, such as higher compression ratio and slower decompression speed than LZ77 (a variant of LZ78). In this paper, we study regular expression matching on LZ77 compressed strings. To address this problem, we propose an efficient algorithm, namely, RELZ, utilizing the positive factors, i.e., the prefix and the suffix, and negative factors (Negative factors are substrings that cannot appear in an answer.) of the regular expression to prune the candidates. For the sake of quickly locating these two kinds of factors on the compressed string without decompression, we design a variant of suffix trie index, called SSLZ. We construct bitmaps for factors of regular expression to detect candidates. Moreover, due to the high space cost of SSLZ, we propose a variant index that partially maintain suffixes of the phrases with high frequency and develop an efficient regular expression algorithm based on the novel index, namely, RELZ+. In addition, two optimization strategies employing block filtering and LZ filtering are proposed to prune false negative candidates. At last, we conduct a comprehensive performance evaluation depending on four real data sets to validate our ideas and the proposed algorithms. The experimental results show that our RELZ and RELZ+ algorithms significantly outperform the existing algorithms.