Towards Markup-Aware Text Compression

Although text compression can be successfully applied to markup languages, it does so without any semantic knowledge of the data types present within the markup. In this paper we illustrate how this added knowledge can be used to develop a hybrid tool which combines traditional text compression with markup-awareness to improve compression size against existing well known text compression tools. Our results show that for highly structured markup it is possible to improve the level of compression by around 20% compared to the best performing existing tool we studied. We describe the limitations of our approach and discuss potential implementation options with the overall goal being to produce a practical Unix-like tool.