Apply the clusters content decisions and fix two clusters parser bugs

LU
3h ago
+199-80

Applies the content-review decisions for the cluster pages: corrections to the Clusters v4.1 Google Doc, regenerated with two parser fixes, and cleaner publication titles in the featured-resources generator.

Clusters Doc and regenerated pages

The Doc was corrected first. Then parse_clusters_to_sheet.py --json-only and generate_cluster_pages.py were re-run.

  • Names: clusters 5, 6 and 8 now use title case, matching the /resources/ filters ("FAIR Data and Materials", "Replication and Meta-Research", "Transparency and Reproducibility in Computation and Analysis").
  • Descriptions:
    • WEIRD now expands "Industrialized".
    • The GOLD and DIAMOND open access glosses are corrected.
    • Rewritten: the equity definition, the LGBTQ+ scope, the communication-barriers list, the Study Design and qualitative-reporting sentences, the Netherlands Code sentence and the cluster 6 grammar.
    • Removed: the repeated NHST clause and the QRP fragment.
    • Rights retention gets an explanatory sentence.
    • Decolonizing Research Practices gets a description.
  • Removed from the Doc: the "INSERT DESCRIPTION" placeholders, an editor's "To add:" note (now the Flynn, 2025 reference) and a Wikipedia note.
  • References:
    • A duplicate Bartoš et al. (2024) is gone.
    • Mozersky et al. (2020) in IASSIST Quarterly now has its own DOI, 10.29173/iq952, checked against Crossref.
    • The URL-encoded DOI 10.1037%2Fqup0000076 is normalised.

Parser fixes (scripts/parse_clusters_to_sheet.py)

  • Descriptions that cite a year: a description line that mentions a year in passing ("Clarke's (2022) piece") stays the description, so it is no longer turned into a reference. Lines that start like an author list, or that carry a DOI or URL, are still read as references. On the earlier Doc text, only the cluster 10 "Standards and guidelines for rigorous qualitative research" sub-cluster changes.
  • Group authors: when the doi.org APA text starts with the year, the Doc's author is kept. This affects Open Science Collaboration (2012, 2015) and National Academies (2017).
  • Tests are added for both fixes.

Publication titles (scripts/build_featured_data.py)

  • Titles and APA references are now stored as plain text plus <i>/<scp> markup. HTML entities are decoded, including double-escaped &lt;em&gt;; <em> becomes <i>; and whitespace is collapsed.
  • This also repairs citation buttons such as "Lakens & amp; Lieck (2022)".
  • A separate PR into rebrand-doors renders the allow-listed tags.
  • The change goes live with the next daily data-processing run.

AGENTS.md

AGENTS.md now points to --json-only. Run on its own, --export-json also clears and rewrites the sheet tabs, including the enriched Publications columns.

Checks: unit tests pass, hugo --gc --minify builds, and check_seo_html.py passes.

🤖 Generated with Claude Code

Summary by cubic

Applies the content-review decisions for the cluster pages and fixes two parser bugs that surfaced when regenerating them: clusters 5, 6, and 8 now use title case, confusing or broken descriptions and references were corrected, and the parser no longer mistakes prose that cites a year for a reference or drops group authors like Open Science Collaboration.

Parser fixes

  • A description line mentioning a year in passing ("Clarke's (2022) piece") stays a description, provided it carries no "(year)." reference date; author-list, DOI, and URL lines are still treated as references.
  • When the doi.org APA text starts with the year, the Doc's group author is kept (Open Science Collaboration, National Academies), including the cluster 8 2015 Science entry.
  • The script's --help, AGENTS.md, and the site's data-error message now recommend --json-only, since plain --export-json also clears and rewrites the Clusters, Sub-Clusters, and Publications sheet tabs.

Content

  • Corrected WEIRD's expansion, the GOLD and DIAMOND open access glosses, equity, LGBTQ+ scope, communication barriers, Study Design, and other cluster 6 descriptions.
  • Removed a duplicate Bartoš et al. reference, the "INSERT DESCRIPTION" placeholders, and the Wikipedia and "To add:" notes; added a description for Decolonizing Research Practices.
  • Publication titles and APA references now decode HTML entities and keep <i>/<scp> markup while leaving all other angle brackets escaped, preserving the full stop in "et al." and dropping it after group authors; this fixes broken citation buttons like "Lakens &amp; Lieck (2022)".