A Library for Research Paper Refs

A personal library of BibTeX references for academic writing.

Naming Convention

Each entry follows the pattern: Author + Topic/Venue + Year

Repository Layout

• bb.bib — aggregated bibliography (generated from the files under refs/).
• refs/ — source .bib files, organized by entry type and topic:
  • 0x-*-example.bib — examples for each entry type (journal, conference, url).
  • 1x-journal-*.bib — journal articles, grouped by topic (authentication, HCI, web, other).
  • 2x-conference-*.bib — conference papers, grouped by the same topics.
  • 3x-url-*.bib — web resources (authentication, HCI, web, general).
  • 4x-techreport-*.bib — technical reports.
  • 5x-degree-*.bib — theses (doctoral / master's).
• scripts/merge_bib.py — utility to merge all files in refs/ into bb.bib.

Editing Rules

• Entries inside refs/ are kept complete — do not trim pages, do not shorten URLs, and do not abbreviate venue names.
• Any trimming or shortening (e.g., dropping pages, replacing a long URL with a short link) should only be done in the per-paper working copy of the bibliography for a specific manuscript, never in this library.
• Conference papers must contain an explicit year = {YYYY} field.
• By local convention, web resources remain @manual entries with URLs in note = {\url{...}}.
• Use BibTeX month macros (jan through dec) instead of mixed month strings.

Usage

Add or update entries in the appropriate file under refs/, then regenerate the combined bibliography:

python scripts/merge_bib.py

The script also accepts custom inputs, output paths, and a filename regex:

# Defaults: merge refs/*-local.bib into bb.bib, then scan for duplicates.
python scripts/merge_bib.py

# Custom output file.
python scripts/merge_bib.py -o all-refs.bib

# Multiple inputs (directories and/or individual .bib files).
python scripts/merge_bib.py refs other_dir refs/extra.bib

# Change which filenames are picked up inside a directory (regex on basename).
python scripts/merge_bib.py -p '\.bib$'

# Skip the duplicate-title scan.
python scripts/merge_bib.py --no-dedup

# Sort each input file's entries by (year asc, key-prefix asc) before merging.
# Sorting is on by default; use --no-sort to keep the original order.
python scripts/merge_bib.py --no-sort

# Same sort, but also rewrite the source files in refs/ in place.
python scripts/merge_bib.py --sort-in-place

# Put newest entries first (or use "key" for cite-key ordering).
python scripts/merge_bib.py --sort-order year-desc

# Preview/apply conservative year and month normalization.
python scripts/normalize_bib.py
python scripts/normalize_bib.py --write

After merging, the script prints:

• Entry keys that don't end with a 4-digit year (likely naming-convention slips).
• Duplicate groups — entries sharing the same @type and a normalized title (lowercased, all non-alphanumeric characters stripped) are reported so you can clean them up by hand. Duplicates are only flagged, never deleted automatically.
• Duplicate cite keys (case-insensitive), duplicate DOIs, and missing core fields.

Sorting details. Entries are ordered by their explicit year/date field, falling back to the trailing cite-key year, then by the full lowercased key. --sort-order supports year-asc, year-desc, and key. Any comment or blank line immediately above an entry travels with that entry, so banner comments like % === 2005 === stay attached to their first entry. Anything before the first @entry is treated as a file header and left untouched.

Cite bb.bib from your LaTeX project, or copy individual entries as needed.

Happy Researching!