Efficient Integer Retrieving from Unordered Compressed Sequences

TL;DR

This paper introduces a method for nearly constant-time direct access to codewords in RMD-coded bitstreams, enhancing data compression efficiency with minimal space overhead, demonstrated on natural language text.

Contribution

It presents a novel technique for direct codeword retrieval from RMD-bitstreams with almost constant time and minimal space overhead.

Findings

Achieves near constant-time codeword access

Maintains good compression ratio in natural language text

Uses minimal additional space for indexing

Abstract

The variable-length Reverse Multi-Delimiter (RMD) codes are known to represent sequences of unbounded and unordered integers. When applied to data compression, they combine a good compression ratio with fast decoding. In this paper, we investigate another property of RMD-codes - the ability of direct access to codewords in the encoded bitstream. We present the method allowing us to extract and decode a codeword from an RMD-bitstream in almost constant time with the tiny space overhead, and make experiments on its application to natural language text compression.