Random access to high-order entropy compressed text

Abstract

This paper is a survey on the problem of storing a string in compressed format, so that (a) the resulting space is close to the high-order empirical entropy of the string, which is a lower bound on the compression achievable with text compressors based on contexts, and (b) constant-time access is still provided to the string as if it was uncompressed. This is obviously better than decompressing (a large portion of) the whole string each time a random access to one of its substrings is needed. A storage scheme that satisfies these requirements can thus replace the trivial explicit representation of a text in any data structure that requires random access to it, alleviating the algorithmic designer from the task of compressing it. © Springer-Verlag 2013.