A collection of small corpuses of interesting data for the creation of bots and similar stuff.
5.1k
Stars
1.3k
Forks
11
Open issues
30
Contributors
AI Analysis
Corpora is a curated collection of small JSON datasets (nouns, adjectives, verbs, etc.) designed for rapid prototyping of bots and creative projects. It serves developers and educators who need accessible sample data for quick iterations and teaching, rather than those seeking exhaustive linguistic resources or production APIs. Best suited for hobbyist bot creators and students learning generative text techniques.
Inferred from signals mentioned in the README (tests, CI, type safety) — not a review of the actual code.
AI's overall editorial judgment — not an average of the bars above, can weigh other factors too.
Curated JSON data collections for bot-makers, educators, and rapid prototypers
Corpora is a public-domain (CC0) repository of small, curated JSON files covering topics like adjectives, nouns, city names, colors, and hundreds of other categories. It was built for creators making Twitter bots, generative art, and similar projects who need clean, ready-to-use data without scraping or cleaning their own sources. It also serves as a teaching resource in workshops on creative coding. Its value is accessibility and breadth, not exhaustiveness. Wrapper packages exist for Python and Node.js, extending its reach across ecosystems.
Created in February 2014 by Darius Kazemi, a prominent figure in the internet art and bot-making community, as a personal utility that grew into a community-maintained shared resource through open pull requests.
Growth was driven organically by the bot-making and creative coding communities of the mid-2010s, particularly around Twitter bot culture. The project accumulated 5,099 stars primarily in its first few years of activity. Growth has since plateaued, consistent with the broader decline of Twitter bot culture after Twitter's API access restrictions in 2023.
The existence of third-party wrappers — corpora-project on NPM, pycorpora on PyPI, and a live Glitch-hosted API — suggests real-world usage beyond casual browsing. The 1,296 forks also indicate active derivative use. Specific download counts or usage metrics are not available in the provided data, so scale of production use is unverified.
Appears to be a flat repository of JSON files organized by category (e.g., words, geography, animals). No application logic — purely static data files. Language is listed as JavaScript likely due to tooling scripts or CI configuration, not core logic.
README mentions CI-based JSON linting via JSONLint for pull requests, suggesting automated validation of data file integrity. No unit tests documented in the traditional sense, as the project is data, not code.
Last push was January 19, 2026 — approximately 5 months before the current date. This indicates the project is still accepting contributions. With only 1 star gained in the last 7 days, growth is minimal but maintenance appears ongoing at a low-cadence pace appropriate for a data repository.
ADOPT IF: you are building generative art, bots, or interactive projects that need small, clean, immediately usable JSON data without infrastructure overhead, or if you are teaching creative coding workshops. AVOID IF: you need exhaustive, up-to-date, or linguistically precise datasets — Corpora is explicitly not meant for that. MONITOR IF: you depend on community contributions to stay current; the contributor pace has slowed and some data may be dated.
Independent dimensions
Mainstream potential
2/10
Technical importance
3/10
Adoption evidence
5/10
- The Twitter bot ecosystem that originally drove adoption has significantly contracted since 2023 API restrictions, reducing the primary use case population.
- Data currency is unverified — some JSON files may reflect outdated facts (e.g., city populations, cultural references) and no systematic update process is documented.
- The license field is listed as unknown in the metadata, though the README clearly states CC0; this metadata inconsistency could cause automated license compliance tools to flag the project.
- Community contribution rate appears to have slowed based on low recent star activity; niche data categories may not receive updates or corrections.
- No formal governance or maintainer succession plan is documented, creating single-point-of-failure risk if the original author becomes inactive.
Corpora will likely persist as a stable, low-activity reference repository used in education and creative coding. It is unlikely to see significant growth without a new wave of bot or generative content culture, but it is also unlikely to become irrelevant given its CC0 license and static nature.
Newsletter
Get analyses like this every Monday
Free weekly digest of the most interesting open-source discoveries.
Languages
Information
- Language
- JavaScript
- Last updated
- 6mo ago
- Created
- 151mo ago
- Analyzed with
- anthropic/claude-haiku-4-5
Stars over time
Contributors over time
Top 100 contributors only — repos with more will plateau at 100.
Open issues
Student Submissions?
[Idea] Enforce 1000 element limit in Github Actions workflow
About Json file format and golang lib support
Capital of Bolivia
Add "feeling cold, warm, cool, hot" in the activities.json file
Top contributors
Recent releases
No releases published yet.
Similar repos
brightmart/nlp_chinese_corpus
A large-scale Chinese NLP corpus repository providing pre-processed datasets...
FurkanGozukara/Stable-Diffusion
This repository is a curated educational resource hub for generative AI,...
| Repository | Stars | Week Δ | Language | Score | Updated |
|---|---|---|---|---|---|
|
dariusk/corpora
this
|
5.1k | +3 | JavaScript | 7/10 | 6mo ago |
 nltk/nltk_data
|
1.8k | — | Python | 7/10 | 1w ago |
|
brightmart/nlp_chinese_corpus
|
9.9k | — | — | 7/10 | 5mo ago |
 withcoral/coral
|
5.1k | — | Rust | 8/10 | 7h ago |
|
FurkanGozukara/Stable-Diffusion
|
2.7k | — | HTML | 6/10 | 2w ago |
 replicate/cog
|
9.4k | — | Go | 8/10 | 9h ago |
Wordnik offers exhaustive linguistic data with metadata; Corpora is intentionally small and offline-first, requiring no API key or network call — different use cases rather than direct competition.
Provides complete, live datasets (e.g., all Wikipedia titles); Corpora provides curated static subsets — better for rapid prototyping, worse for completeness.
A large-scale NLP corpus for Chinese language tasks; targets ML researchers, not creative coders. Entirely different audience and scale.
Faker libraries generate random fake data programmatically; Corpora provides curated human-selected lists that can be used similarly but are transparent, inspectable, and domain-specific.
Hugging Face hosts datasets at much larger scale for ML pipelines; Corpora targets creative prototyping with minimal setup and no ML infrastructure dependency.
