emsenn

Bibliography automation

2 min read

Summary

Generate bibliography.bib on demand from references: frontmatter fields distributed across the vault. The per-file references: field is the source of truth for bibliographic data; the consolidated .bib file is a derived artifact.

Motivation

A hand-maintained bibliography is fragile and disconnects from the content that uses it. The citation spec (semiotic-markdown/citations.md) now specifies that references: frontmatter fields carry structured bibliographic data per-file, and that bibliography.bib SHOULD be generated from these fields. This plan implements that specification.

Each file’s references: field makes the file self-standing — a reader does not need the centralized .bib to understand what is cited. The generated .bib file serves build-time tools (pandoc-citeproc) and cross-vault aggregation.

Steps

  1. Write generate-bibliography.py alongside existing scripts in technology/specifications/agential-semioverse-repository/scripts/. Reuse the parse_frontmatter function from predicate-graph.py.
  2. The script should:
    • Scan all .md files for references: frontmatter fields
    • Extract structured bibliographic data (authors, title, year, etc.)
    • Detect and report conflicts (same citekey, different data)
    • Generate bibliography.bib in BibTeX format
    • Report audit findings: orphaned [@citekey] references, unresolvable cites: entries, inconsistent duplicates
  3. Write a companion SKILL.md for the generation/audit.
  4. Migrate existing bibliography.bib entries into references: fields on the files that use them (incremental, highest-cited first).
  5. Migrate free-text cites: entries to resolvable format where possible (citekey or structured “Author. Title. Year.” format).

Done when

  • generate-bibliography.py compiles references: into valid BibTeX
  • Audit mode reports orphaned citekeys and unresolvable cites
  • SKILL.md exists for generation and audit
  • Top 20 most-cited sources have references: entries on their files
  • bibliography.bib matches generated output (or is replaced by it)

Dependencies

  • predicate-graph.py (for frontmatter parsing infrastructure)
  • Citation spec (semiotic-markdown/citations.md) for field definitions

Log

2026-03-07 — Created. Prompted by emsenn observing that hand-coded bibliography.bib is fragile and disconnected from content.

2026-03-07 — Revised. Direction clarified: references: frontmatter is the source of truth, bibliography.bib is a generated artifact. Citation spec (semiotic-markdown/citations.md) updated to specify on-demand bibliography generation, resolvable cites: formats, and audit capabilities. The open question (per-file vs centralized) is resolved: per-file references: is the source of truth; centralized .bib is generated from it.

Graph View