NOTICE

LOL
Copyright 2024-2026 Jonathan D.A. Jewell and Contributors

This project contains data files derived from 1000Langs
(https://github.com/ehsanasgari/1000Langs), Copyright (c) 2018-2019
Ehsaneddin Asgari, licensed under the Apache License, Version 2.0. A copy of
that licence is in LICENSES/Apache-2.0.txt. 1000Langs has no NOTICE file of
its own, so there is nothing further to carry over.

What was taken
--------------
Every file under meta/ (8 data files and 196 flag images) is an unmodified
copy of the same path in 1000Langs. They are still under Apache-2.0. Some of
them are material that 1000Langs itself redistributed: captured web pages
(meta/bibles_list.html, meta/ebible.html), an API response
(meta/api_volumes.txt), a language table (meta/ISO639-3.tsv) and flag images
(meta/flags/). Their original authors are not recorded in 1000Langs and are
not asserted here. REUSE.toml lists the licensing file by file.

What was not taken
------------------
LOL's code contains no code from 1000Langs. It is an independent
reimplementation of the same idea, written first in ReScript and now in
AffineScript, rather than translated from the Python. The significant
differences are:

  * Language and runtime: AffineScript/ReScript on a JS runtime, plus an
    Idris2 ABI, Zig FFI, Agda proofs, a Julia analysis layer and Nickel
    configuration, in place of Python with requests, lxml and pandas.
  * APIs: API.Bible v1 in place of the Digital Bible Platform v2 (dbt.io).
  * Crawling: endpoint-based fetches per chapter with a rate limiter and
    retry policy, in place of following "next" links with XPath.
  * Data model: typed verse and chapter records in place of BBBCCCVVV-keyed
    TSV files.
  * Sources: eBible.org and Find.Bible are crawled for text; 1000Langs used
    them only for metadata.
  * Added: a VeriSimDB corpus-quality scan and an OpenCyc integration stub.
  * Removed: the Massive Parallel corpus comparison, per-source crawl reports,
    repeat-for-coverage loops and metadata refresh.

The evidence for this split is in docs/provenance/1000langs-comparison.adoc.