Grammar compression with probabilistic context-free grammar
Abstract
We propose a new approach for universal lossless text compression, based on grammar compression. In the literature, a target string has been compressed as a context-free grammar in Chomsky normal form satisfying . Such a grammar is often called a \emph{straight-line program} (SLP). In this paper, we consider a probabilistic grammar that generates , but not necessarily as a unique element of . In order to recover the original text unambiguously, we keep both the grammar and the derivation tree of from the start symbol in , in compressed form. We show some simple evidence that our proposal is indeed more efficient than SLPs for certain texts, both from theoretical and practical points of view.
Keywords
Cite
@article{arxiv.2003.08097,
title = {Grammar compression with probabilistic context-free grammar},
author = {Hiroaki Naganuma and Diptarama Hendrian and Ryo Yoshinaka and Ayumi Shinohara and Naoki Kobayashi},
journal= {arXiv preprint arXiv:2003.08097},
year = {2020}
}
Comments
11 pages, 3 figures, accepted for poster presentation at DCC 2020