Four-Layer Grapheme Model for Computational Paleography

Abstract

This article proposes a novel mathematical model of logical relationship among glyphs belonging to the same grapheme. Its research belongs to the computational paleography that is a field in the applied computer science. The proposed grapheme model is presented in four logical layers from bottom to up namely as Topology, Visual Identity, Phonetic, and Semantic Layer. In the Topology Layer, a unique glyph is defined by a set of topological properties. When trying to describe the logical relation of various glyphs, their topological properties must be examined in a higher layer framework so called Visual Identity Layer. In that layer, the glyphs of a single grapheme share some topological attributes in common. These common topological attributes form a main identity of a grapheme, which is called Common Identity template that is obtained by means of Supervised Learning method. The Phonetic Layer gives the sound values associated to the grapheme, and the Semantic Layer describes the usage of the grapheme in texts. Some potential implementations of the grapheme model are also presented.