Automata on Lempel-Ziv compressed strings

Abstract

Using the Lempel-Ziv-78 compression algorithm to compress a string yields a dictionary of substrings, i.e. an edge-labelled tree with an order-compatible enumeration, here called an LZ-trie. Queries about strings translate to queries about LZ-tries and hence can in principle be answered without decompression. We compare notions of automata accepting LZ-tries and consider the relation between acceptable and MSO-definable classes of LZ-tries. It turns out that regular properties of strings can be checked efficiently on compressed strings by LZ-trie automata. © Springer-Verlag Berlin Heidelberg 2003.