LoginSign Up
← Back to all sparks
O

ojoregex

ANALYTICS
Velocity0.0

R package ojoregex by openjusticeok — release notes from GitHub.

github.com/openjusticeok/ojoregex ↗

Oklahoma's court-data nonprofit maintains the regex layer that turns charge text into categories.

criminal-justicecourt-datar-packagetext-classificationregexcivic-tech
◆Current state
ojoregex is Open Justice Oklahoma's pattern library for classifying criminal charge descriptions from court records — the unglamorous translation layer between free-text charge fields and analysable categories. Its entire release history reached this feed as four tags published within three minutes, so the feed order reflects a backfill rather than a shipping cadence. Release notes are merge references rather than descriptions, which limits how much can be read from the changelog alone.
◆Where it's heading
What the notes do show is a package alternating between domain corrections and R tooling upkeep: a fix to property-crime matching in one release, dplyr select semantics in the next. That is the expected shape for a regex catalogue — accuracy work arrives as individual charge types get miscategorised in real analyses, and the rest is keeping the package installable against a moving tidyverse. Contributions come from a small internal team, and the vignette work referenced in the earliest tag suggests the pattern list doubles as documentation for analysts.
◆Prediction
The visible pattern is incremental match fixes as charge types surface in use; the release notes carry too little detail to predict anything beyond that without reading the underlying pull requests.

◆Recent moves

  1. 1mo ago

    Patch for dplyr select semantics

    A single-PR patch keeping the package working against current dplyr selection behaviour. Tidyverse upkeep, not a change to any pattern.

    View source ↗
  2. 1mo ago

    ojoregex v0.10.0

    A minor version bump whose release notes consist of a single merge from the development branch, with no description of what changed. Nothing in the feed indicates whether patterns, interfaces, or internals moved — the direction of this release simply is not readable from the changelog.

    View source ↗
  3. 1mo ago

    Namespace prefixes fixed, plus a property-crime match bug

    The clearest instance of this package's core work: a specific charge type — malicious injury to property — was matching incorrectly, and the fix ships alongside namespace prefix corrections. Accuracy on individual charge descriptions is the whole value proposition of a classification regex library, so a miscategorised offence is a data quality bug for every analysis downstream.

    View source ↗
  4. 1mo ago

    First tagged release, carrying the whole development history

    The earliest tag in the repository, so its notes roll up every pull request that preceded it — property crime patterns, edge case fixes, an appendix vignette, and a 0.1.0 milestone described as incomplete but usable for analysts. It reads as a project that was already in production use internally before it acquired version numbers.

    View source ↗