Skip to content

About

Academic website of Brian W. Locke, MD, MSCI—pulmonary and critical care physician-scientist studying respiratory failure, clinical data, prediction, and pragmatic trials.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Latest commit

 

History

476 Commits

Folders and files

Repository files navigation

Brian W. Locke academic website

This repository publishes https://reblocke.github.io/ as a small static Jekyll site and maintains its public scholarly metadata.

Routine edits

Human-maintained facts live in only three files:

  • _data/person.yml: identity, affiliations, profiles, portrait, and disclosure.
  • _data/cv.yml: appointments, training, support, teaching, service, and honors.
  • _data/work.yml: curated publications, presentations, topic collections, repositories, live project-site links, relationships, and editorial selections.

Work items distinguish publications (journal-article, conference-paper, preprint, commentary, letter), presentations (abstract), and scholarly products (protocol, dataset, software, educational-resource). Only publication types enter /publications/; selected products have separately labeled sections on /cv/. Optional url, quoted ISO date, version, and status describe the cited record without implying peer review, clinical validation, or a formal software release. Preserve each product's own author order. Use related_ids to link companions and repositories; a repository paired through display_with stays in the catalog but does not repeat as standalone Work. Keep a DOI on its publication/product record and resolve the repository's citation through related_id.

Files under _data/external/, _data/generated/, llms.txt, and research-repositories.* are generated. Do not edit them manually.

Local development

Install Ruby 3.3.12 and Bundler, then run:

bundle install
bin/check
bundle exec jekyll serve

The local site is available at http://127.0.0.1:4000/.

To regenerate deterministic outputs after editing canonical data:

bundle exec ruby scripts/generate_indexes.rb --write
bin/check

bin/check is deterministic and uses only repository-local inputs. It validates source data, regenerability, the built site, canonical indexability, structured-data graphs, rendered publication/topic completeness, and negative mutations that must be rejected. The built-site validator and mutation suite also accept an alternate artifact through --site PATH.

The separate Repository readiness audit workflow inventories public repository documentation monthly and on demand. It resolves one commit per repository and reports file access, exact README casing, placeholders and structural term matches. Its result is not semantic documentation acceptance, scientific reproduction, or release readiness; semantic review is reported as not_evaluated. An absent license or agent-instruction file is an observation unless an applicable project policy requires it. To run the advisory inventory locally with GitHub authentication:

GH_TOKEN="$(gh auth token)" python3 scripts/audit_llm_readiness.py --manifest research-repositories.csv --advisory

External metadata refresh

The monthly workflow retrieves public metadata from GitHub, ORCID, Crossref, and PubMed, regenerates the site, validates it, and opens a review pull request. It never publishes directly.

The generated reconciliation report separates unmatched ORCID candidates, fields missing from canonical records, and material source conflicts. DOI or PMID matches anywhere in the canonical work registry count as represented; external metadata never overwrites curated content automatically.

The refresh retains the website repository's public GitHub metadata but omits its self-changing updated_at value so deployment commits do not manufacture metadata-only pull requests.

Run the refresh locally with:

GITHUB_TOKEN="$(gh auth token)" bundle exec ruby scripts/refresh_external_metadata.rb
bundle exec ruby scripts/generate_indexes.rb --write
bin/check

ORCID_CLIENT_ID and ORCID_CLIENT_SECRET may be configured for authenticated ORCID Public API access. The public endpoint is used for this single public record when credentials are absent. NCBI_API_KEY is optional for higher PubMed request limits.

Google Scholar remains the linked comprehensive profile. It does not provide a supported automated profile API, so the site does not scrape it.

Publishing

Pull requests and master run the same bin/check build. A push to master deploys the exact validated _site artifact through GitHub Pages Actions. Repository Settings → Pages → Build and deployment must use GitHub Actions.

Search discovery and Search Console

The generated sitemap.xml contains only the six canonical routes: /, /work/, /publications/, /topics/hypercapnic-respiratory-failure/, /cv/, and /research-repositories/. The former /bio/ route redirects to the homepage About section. The generated robots.txt allows crawling and advertises https://reblocke.github.io/sitemap.xml. Canonical and historical redirect pages must not carry blocking robots, googlebot, or bingbot directives. The 404 and Notice pages are intentionally excluded with noindex,follow and omitted from the sitemap.

Each canonical page has a unique search title and description, a self-canonical URL, at least one crawlable internal link, and one schema.org @graph grounded in the canonical person data. bin/check verifies those contracts as well as rendered publication/product placement, citation metadata, related links, and the curated respiratory topic. These controls support ordinary crawler discovery without a Google account, but they do not provide Search Console reports or prove that Google has indexed a URL.

The URL-prefix property https://reblocke.github.io/ uses a public HTML verification tag sourced from google_site_verification in _config.yml and rendered only in the homepage <head>. Retain the tag after verification because Google checks it periodically. If Google issues a replacement, update it through the normal pull-request and validated Pages deployment workflow.

A signed-in maintainer completes the Google-side handoff from Google Search Console:

  1. Confirm that the live homepage contains the exact verification tag, then select Verify for the URL-prefix property.
  2. In Sitemaps, submit sitemap.xml.
  3. Use URL inspection for the six canonical routes listed above and request indexing when useful.

The HTML verification tag is intentionally public. Google passwords, OAuth authorization codes, access or refresh tokens, recovery codes, and browser-session data are secrets and must never be committed or shared for this workflow. Sitemap submission and indexing requests are asynchronous and do not guarantee immediate indexing.

Public-content boundary

Do not commit PHI, restricted datasets, credentials, private repository names, collaborator-only drafts, unpublished grant material, or publisher-formatted PDFs. Link to DOI, PubMed, PubMed Central, accepted manuscripts with verified rights, and public repositories.

Project-site links are an editorial allowlist, not an automatic mirror of every GitHub Pages deployment. Add paired live_url and live_label fields only for a selected public repository whose site is stable, useful to visitors, and safe to surface. The generator carries those fields into JSON and CSV, and the rendered Work and repository-catalog pages must both link to each curated site. Individual project repositories retain ownership of their own crawling, metadata, accessibility, and release policy.

License and attribution

See LICENSE and NOTICE.md.

About

Academic website of Brian W. Locke, MD, MSCI—pulmonary and critical care physician-scientist studying respiratory failure, clinical data, prediction, and pragmatic trials.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages