brawer / osmdiffs / 42315762001

94%

push

github

feat(matchers): Match features on their websites (#896)

* Match ATP and OSM features on website URLs

Website, contact:website and url values are normalized and hashed like
phone numbers and email addresses, and count as contact evidence.
Websites shared by unrelated ATP features go to a third nondescript
set, alltheplaces.nondescript-websites. A website alone doesn't count
for the coverage of deletion candidates.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ttfb2MGwS3XXMu6tBsrXLX

* Treat websites shared over 100 m as nondescript

A web page that ATP lists for places a few hundred meters apart, such
as one page for all kindergartens of a Zürich school district,
describes a group, not a place. Phones and emails keep the 1 km spread,
since same-site sharing is common for them.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ttfb2MGwS3XXMu6tBsrXLX

* Add website-matching note and eval regions to brand-domain note

Both notes now report every eval region, as the baseline for later
changes.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ttfb2MGwS3XXMu6tBsrXLX

---------

Co-authored-by: Claude <noreply@anthropic.com>

109 of 111 new or added lines in 5 files covered. (98.2%)

15484 of 16351 relevant lines covered (94.7%)

4007.18 hits per line