docs(backlog): record relevance.ts left-word-boundary false positives

[ltanafranca1004]

What & why

Records a real defect found during a manual review of #96/#97, both of which merged under an explicit no-CodeRabbit-review exception when the free-tier review quota ran out.

Worth stating up front: a retroactive CodeRabbit review of those two PRs is not obtainable. I tested it — @coderabbitai review returns ⚠️ Already reviewed on #96 (its incremental system considers the earlier commit reviewed) and ⚠️ Pull request is closed on #97. Not rate limits; the quota had cleared. So a manual pass was the substitute, and this entry is what it produced.

Docs only. No code change.

The finding

Same bug class CodeRabbit itself caught on #96 — tech → \btech, which was matching the tail of "Biotech Help" — but in terms that predate that PR. A sweep of all 67 terms in RELEVANCE_RE found 60 lack a leading \b. Most are harmless (no English word ends in "newcomer"). Four are substrings of common words:

term	swallowed by	real-looking title that wrongly passes
lease\b	P·lease	Please Note: Library Closed Monday
rental	Pa·rental	Parental Controls: Keeping Kids Safe Online
tenant	Lieu·tenant	Lieutenant Governor Reading Award Ceremony
resume	P·resume·d	Presumed Innocent: Film Screening

Each verified against the live filter, not reasoned about on paper.

Status: latent, zero live incidence

please, parental, lieutenant, presum and resumed each return 0 matches across:

• the 395-title captured corpus (all seven filtered sources' full 4-month windows), and
• all 206 rows currently in public.events

Nothing is being wrongly ingested today. Parental … is the likeliest to fire eventually since it's ordinary library programming — so worth fixing before it does, but not urgent.

Why recorded rather than fixed here

Adding \b narrows the regex. The monotonicity argument that makes term additions safe ("a union alternative can only add keeps") does not apply — a narrowing edit can drop an existing keep. It needs the same corpus re-verification the \btech fix got, plus DROP fixtures in lib/relevance_test.ts.

Best folded into whichever PR next touches relevance.ts — e.g. the near-misses already recorded directly above this entry — rather than shipped as its own change.

Placement

Deliberately a separate entry rather than folded into the existing relevance.ts near-misses section. That section is about terms the filter wrongly drops (false negatives); this is the opposite direction — titles it wrongly keeps. Different failure mode, different fix shape. Placed adjacent so they read together.

Verification

• npx tsc --noEmit — clean
• deno test supabase/functions/events-crawler/lib/relevance_test.ts — 2/2 pass
• git status — BACKLOG.md is the only file touched

🤖 Generated with Claude Code

Summary by CodeRabbit

• Documentation
  • Added a backlog finding documenting potential false-positive matches in event relevance detection.
  • Recorded that current corpus and live-event checks found no affected incidents.
  • Proposed tighter matching rules and regression coverage for future validation.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
web-app	Ready	Preview	Aug 9, 2026 6:21am

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: d064d1c3-4277-4d8b-91d8-43bf2d0177b6

📥 Commits

Reviewing files that changed from the base of the PR and between 94d9c24 and 3bbebbf.

📒 Files selected for processing (1)

• BACKLOG.md

---

Walkthrough

The change adds a backlog finding for four events-crawler relevance terms that can match substrings within unrelated words. It records zero observed incidents and specifies regex, regression-test, and corpus-validation follow-up work.

Changes

Events crawler relevance audit

Layer / File(s)	Summary
Relevance term audit BACKLOG.md	Documents four relevance terms without leading word boundaries, affected examples, verified zero incidence, proposed regex updates, regression fixtures, and corpus validation requirements.

Estimated code review effort: 1 (Trivial) | ~3 minutes

Possibly related PRs

• UnifyCN/web-app#96: Addresses related events-crawler relevance regex boundaries and regression coverage.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and concisely describes the documentation change and identifies the recorded relevance false positives.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch docs/relevance-word-boundary-finding

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[ltanafranca1004]

@coderabbitai review

[coderabbitai]

✅ Action performed

Review finished.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.