iter 1: preserve date qualifiers in web entries
First real loop iteration found that GPT-5 silently drops date qualifiers
("Effective", "Published", "Accessed", etc.) when reformatting web
sources. The field diff was satisfied because the canonical date field
did not include the qualifier, but the canary exact-match axis caught
the regression.
Two fixes, per the canary-upstream policy:
1. Tighten the web-page exemplar canonical: date field now includes the
"Effective" qualifier so the field diff will catch future regressions
without relying on the canary.
2. Add SYSTEM_PROMPT rule 12 instructing the formatter to preserve
semantic date qualifiers from the input.
After these fixes: scalar 1.000, canary exact-match 1.000 on all three
seed exemplars.
This commit is contained in:
@@ -12,5 +12,9 @@ expected_bibliography = "Google. \"Privacy Policy.\" Privacy & Terms. Effective
|
|||||||
author = "Google"
|
author = "Google"
|
||||||
title = "Privacy Policy"
|
title = "Privacy Policy"
|
||||||
site = "Privacy & Terms"
|
site = "Privacy & Terms"
|
||||||
date = "November 15, 2023"
|
# Include the date qualifier ("Effective", "Published", "Accessed", etc.) in
|
||||||
|
# the canonical date field. CMOS 18 preserves these qualifiers in web-source
|
||||||
|
# entries; dropping them is a real regression even though the resulting
|
||||||
|
# string is structurally well-formed.
|
||||||
|
date = "Effective November 15, 2023"
|
||||||
url = "https://policies.google.com/privacy"
|
url = "https://policies.google.com/privacy"
|
||||||
|
|||||||
@@ -73,6 +73,14 @@ Rules you must follow (CMOS 18th edition specifically):
|
|||||||
11. Web page format: Author-or-Organization. "Page Title." Site Name.
|
11. Web page format: Author-or-Organization. "Page Title." Site Name.
|
||||||
Effective/Published Date. URL.
|
Effective/Published Date. URL.
|
||||||
|
|
||||||
|
12. PRESERVE date qualifiers from the input. If the source provides a
|
||||||
|
qualified date such as "Effective November 15, 2023", "Published March
|
||||||
|
3, 2024", "Accessed October 1, 2022", "Posted June 7", "Last modified
|
||||||
|
...", or "Updated ...", keep the qualifier word in the output — do not
|
||||||
|
silently drop it. The qualifier is semantically meaningful: "Effective"
|
||||||
|
dates a policy's activation, "Accessed" marks when the researcher saw
|
||||||
|
the page, etc. CMOS 18 preserves these distinctions.
|
||||||
|
|
||||||
Output ONLY the single reformatted entry. No preamble, no explanation.
|
Output ONLY the single reformatted entry. No preamble, no explanation.
|
||||||
"""
|
"""
|
||||||
|
|
||||||
|
|||||||
Reference in New Issue
Block a user