Fundamental concepts

This section explains what makes Tabulae different from other approaches to Latin morphological parsing.

Key points

• Tabulae datasets are composed of tables of inflectional rules, and tables of vocabulary stems.
• Each dataset follows a specified orthography.
• Forms and lexemes are identified by URNs that are independent of orthography.
• Datasets are compatible between the Scala and Julia implementations of Tabulae.
• Data tables use abbreviated URNs; adding a URN registry to the dataset allows expanding abbreviated URNs to full Cite2Urns.
• Tabulae.jl is a pure Julia system: it has no external dependencies
• Tabulae.jl implements the model of citable morphological parsing defined the CitableParserBuilders package (documentation)

Architecture

Datasets

• datasets: data can be maintained in directories of delimited-text files for stem and rules tables
• rule types and stem types
• reading/writing datasets uses fromcex and cex functions of the CitableBase package

Building parser

• Cartesian product of rules * stems, filtered on inflectional class