There's a lot of open issues
A long-lived project that still receives updates
Map data package for Interscript interoperable script conversion systems.
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
 Dependencies
 Project Readme

CI status = Interscript: Script conversion maps

Introduction

This repository is the single source of truth for Interscript’s transliteration corpus: 300 authority-backed script conversion systems encoded in the "Interoperable Script Conversion" language (.isc).

Interscript turns the world’s transliteration standards from paper tables into executable, verifiable software: authority schemes (BGN/PCGN, ISO, UN/UNGEGN, ALA-LC, ICAO, ODNI, …​) are encoded once here, and every runtime — Ruby, TypeScript, Python, the public API, and the website — produces provably identical output.

Contents

Systems from these authorities are included in this repository:

  • ALA-LC

  • BGN/PCGN

  • ICAO

  • ISO

  • ODNI

  • UN (by UNGEGN)

  • and other systems that do not have a defined authority

Organization

All maps are encoded in the "Interoperable Script Conversion" language syntax (extension: .isc).

System file names adhere to the script conversion system code convention of ISO 24229.

Quality gates

CI runs on every change:

  • ISC parse validation in Ruby and TypeScript;

  • each map’s own embedded tests;

  • cross-runtime parity (byte-exact agreement between the Ruby interpreter and the TypeScript runtime over the test corpus);

  • the published artifact (interscript-maps gem and the maps release artifact) is built from this corpus by CI, never by hand.

Coverage and known gaps

Every authority document the project holds has been audited against the corpus. The complete reference-to-corpus inventory — including which standard produced which map, and what remains — is docs/coverage/README.adoc.

Held standards not yet encoded

Fifteen deterministic tables are held but not yet encoded. Each is a complete table with its own example rows; none needs training data to encode:

  • ODNI / ICS 630-01 — Kurdish (Annex G: Perso-Arabic to Latin for personal names, no diacritics) and Chinese (Annex D: not a Han-character romanizer, but an IC standard for names already written in Hanyu Pinyin, with Wade-Giles and Yale conversion appendices).

  • DIN — 31635 (Arabic, Ottoman Turkish, Persian, Kurdish, Urdu), 31636 (Hebrew), 32706 (Armenian), 32708 (Japanese), and 1460-2 (non-Slavic Cyrillic).

  • BSI — BS 2979:1958 (Cyrillic and Greek) and PD 6505:1982 (Korean romanization).

  • BGN/PCGN — Burmese, Hebrew, Khmer, Lao, Shan, and Inuktitut.

Not doing: neural models for five languages

Burmese, Dzongkha, Inuktitut, Lao, and Shan have complete BGN/PCGN tables, but no gold script-to-romanization data exists to train a neural model. Interscript withholds a model rather than ship a bad-quality one; the deterministic maps are still in scope.

Source documents not held

Some verified maps have no archived source document in the project: most ALA-LC tables (one table document is held of the 78 in the catalogue), most ISO standards (only ISO 9, ISO 9984 and ISO/TR 11941 are held), and the UN/UNGEGN schedules. These are archival gaps, not encoding gaps.

Contributing

New or corrected systems are welcome as .isc changes — see CONTRIBUTING.md and the Map Editing Guide in the project monorepo. Alias/dependency declarations must be explicit; generated artifacts are never edited by hand.

This is a Ribose project. Copyright Ribose.