A method for language attribution based on assessment of text irregularity

Abstract

In the work, it is proposed to use a fractal driven index irregularity as a base for language attribution of text. A method for such an attribution of electronic texts is offered. The method may be applied in Big Data analysis, electronic library operation, natural language processing and so on. We examined texts in nine languages, set up with two major fonts. The approach shows the ability to distinguish different languages using the offered irregularity based index only, without reading texts and expert language assessment. The method may be further extended for texts in bitmap. © 2018 Author(s).