Corpora is a collection of small JSON datasets designed for rapid prototyping of bots and generative projects.
The project addresses the friction of repeatedly hunting for, cleaning, and parsing data when building experimental software. Rather than requiring exhaustive sources, Corpora provides curated, reasonably-sized lists—typically around one thousand items—of things like adjectives, nouns, verbs, and other categorical data. Each corpus is stored as a JSON file, making it language-agnostic and immediately usable by any tool that can parse JSON. This approach lets developers quickly test ideas without the overhead of data acquisition, then swap in more comprehensive sources once a prototype proves viable.
Corpora suits developers building bots, generative art, or other experimental internet projects where speed of iteration matters more than completeness. It works well as a teaching tool when time constraints prevent students from sourcing their own data. The project explicitly positions itself as a starting point rather than a replacement for exhaustive APIs like Wordnik or the MediaWiki API. If your project needs every word in the English language or complete metadata, you should look elsewhere. If you need a quick, usable sample to validate an idea or teach a concept, Corpora provides exactly that scope.
The project accepts pull requests for new datasets and maintains a language-neutral approach to ensure data remains accessible regardless of implementation language. Related tools exist to wrap the data—including Node.js packages, a Python interface, and a JSON API—but the core repository remains a collection of static data files rather than code.
