Skip to main content

Publications & Bibliography

Created: 14-06-2026 · Last updated: 07-07-2026

How to cite CDSL

See Abbreviations & Citations for the project's official in-text and bibliographic citation form, plus the recommended wording for acknowledging CDSL data reuse in a website or application.

About the project and its methods

  • Kapp, Dieter B., & Thomas Malten. Report on the Cologne Sanskrit Dictionary Project. Describes how the printed Monier-Williams dictionary was encoded without altering its complex structure or losing information. PDF · Semantic Scholar
  • Funderburk, Jim, & Thomas Malten. "Marking Monier" (May 2008), Second International Sanskrit Computational Linguistics Symposium. Walks through the four phases of digitizing and marking up the Monier-Williams dictionary. HTML

Ongoing record

  • csl-newsletter — a daywise log of project activity (September 2021 onward), kept "so that an academic paper may be published later, if desired or required." See also the History timeline.
note

This is a starting bibliography of the project's own and closely related publications — not an exhaustive list of the scholarship that uses CDSL data. Suggest additions via the GitHub repo.

Publishing about CDSL — ACL venues and standards

The computational-Sanskrit world publishes inside the ACL Anthology — a paper archive, not a data repository: work "gets into the Anthology" by appearing at an indexed venue. The venues below were liveness-verified 07-07-2026 (survey: ACL_DH_COMPATIBILITY_ANALYSIS.md; re-verify any deadline before acting on it later):

VenueAnthology-indexedCadence / next editionFit for CDSL work
World Sanskrit Conference — §23 Computational Sanskrit & DH✅ since 202520th WSC, Dec 2027, IIT BombayBest fit — the exact audience
ISCLSbiennial, ~2028Strong — Sanskrit lexical resources are core scope
LaTeCH-CLfL (SIGHUM)annual — 2027 CFP expected ~Nov–Dec 2026Nearest actionable deadline; historical-language resources are core scope
NLP4DHannualGood; slightly more tool/method-oriented
WILDRE (at LREC)biennial with LREC, ~2028Strong — Indic-resource-specific
LRECbiennial, 2028Flagship language-resource venue

For a pure data paper (no research claims), the non-ACL Journal of Open Humanities Data is the standard DH outlet — but it presupposes an openly deposited, DOI-bearing dataset.

What an ACL-family submission must have (per the ACL style files and the ARR CFP, in force 2026):

  1. Structure: abstract → related work → data → method → evaluation → error analysis.
  2. A mandatory "Limitations" section — unnumbered, before references, outside the page limit; missing it is desk rejection.
  3. The Responsible NLP checklist, whose §B requires citation, license, provenance, and use-restriction disclosure for every artifact used or released — this site's data cards page exists so that section can be filled honestly for anything built on these feeds.
  4. Evaluation to community standard: named metrics (P/R/F1, accuracy, MAP, Cohen's κ for agreement), an explicit baseline, and uncertainty on small-n results — see the metric convention below.
  5. Long papers 8 pages content / short 4; anonymized review.

Presenting results — the metric convention on this site

Any measured result surfaced on this site (a hypothesis test, a coverage figure, a benchmark score) is presented the way an ACL reviewer expects, so a page section can move into a paper without re-derivation:

  • metric + n + baseline + uncertainty + artifact, in that spirit: name the metric, give the sample size, compare against an explicit chance/majority baseline, put a 95% Wilson interval on small-n proportions and a significance level on correlations, and link the committed artifact + script that reproduce the number.
  • Worked examples: the Guides Hypotheses page (GH-1's accuracy with Wilson CI and two baselines; GH-2's ρ with n and p) and the shared-task leaderboard (scorer prints the interval).
  • Agreement between annotation passes is reported as Cohen's κ, not raw percent — and where only a single pass exists yet (the routing gold panel), that absence is stated rather than papered over.
note

Venue choice for any concrete CDSL paper is a human decision (tracked in the project's own planning, not on this page); this section documents the landscape and the bar.

Dr. Mārcis Gasūns