Machine readable data format for recommendations

celestehorgan commented 3 years ago

The problem

Define a machine-readable data format for language recommendations so that recommendations can be used in linting, AI, and other applications.

This task should also include

Requirements

The format must be a common data format (yaml, json, etc).
It must be easy to maintain

Other considerations:

What are the optional/non-optional fields?
How do you validate that a recommendation is in the correct format?
How we ensure that recommendations are kept consistent between the website & the data format? What is the 'single source of truth'? (Suggestion: keep the recommendations in a spec-like format in their own repo, and use Hugo data templates to read recommendations into the website, rather than duplicating)

Blockers

The Language WS needs to produce at least 3 recommendations to work off of.

markcmiller86 commented 3 years ago

We should probably consider taking our inspiration for format(s) from other similar technologies such as translators, spell-checkers, syntax checkers, linters -- none of those really represent a perfect match for the kinds of replacements or substitutions we need in INI but they are ascii/text processing adjacent.

gkarthiks commented 3 years ago

Use case:

From the developer standpoint, for example, I have a GitHub Action that is expected to be used for flagging the issues that consist the non-inclusive language.

From this scenario, I would expect to have machine-readable data that has the non-inclusive language and set of recommendations to it.

Link to the Actions Repo: https://github.com/gkarthiks/nil-issue-reporter

inclusivenaming / org