Abstract

Adaptive context modeling has emerged as one of the most promising new approaches to compressing text. A finite-context model is a probabilistic model that uses the context in which input symbols occur (generally a few preceding characters) to determine the number of bits used to code these symbols. We provide an introduction to context modeling and recent research results that incorporate the concept of context modeling into practical data compression algorithms. 1. Introduction One of the more important developments in the study of data compression is the modern paradigm first presented by Rissanen and Langdon [RL81]. This paradigm divides the process of compression into two separate components: modeling and coding. A model is a representation of the source that generates the data being compressed. Modeling is the process of constructing this representation. Coding entails mapping the modeler's representation of the source into a compressed representation. The coding component tak...

Citations