googleartsculture / fabricius-language-schema Goto Github PK

I'm assuming that the symbolic layer refers to the assignment of glyph labels to areas of an image. In hieroglyphic writings, gaps may be generated by erosion, etc. This creates uncertainty when assigning glyph labels based on absent or incomplete image data (see TLA dictionaries). Contextual data (e.g. adjacent symbols, and putative semantics) may be required to fill the gaps, and to generate a probability distribution over a set of candidate glyphs. Parsing a symbolic sequence into words might assist with this process because gaps can be localised and context limited.