3.0.1

Fixes an issue where the published npm package did not match the source code. Please see the release notes for 3.0.0.

3.0.0

Deprecated - Please use 3.0.1 or later.

• Added more fine grained error messages for validating labels and description.
• Added validation to stop adding of labels that may overwrite standard object properties such as constructor or __proto__.
• Added normalization of descriptions and diacritic processing.
• JSDoc and typescript type information have been added to aid development with whichx.

This release requires a major version release because the changes are backward incompatible with earlier versions. If you are using the import feature, your model will need to be retrained otherwise classification of text with diacritics may fail. Additionally, any labels that clash with existing object properties will not longer be accepted when they may have been earlier.

2.2.0

Added the ability to change the default stop words used by WhichX so that it can be used with more languages without forking.

2.1.3

Fixed a bug where classify would return the default value when all the words were against a single label.

2.1.2

Fix bug where node import failed.

2.1.1

Security patch upgrades of development dependencies.

2.1.0

• Add export function to save a model of what the classifier has already learned.
• Add an import function for importing a previously saved model to a new WhichX object.

2.0.0

• Rename project to the more general title 'WhichX'.
• Update developer dependencies with security vulnerabilities.

1.0.1

• Added unit tests.
• Added linting.
• Published to NPM.
• Lowered m estimate to avoid avoid over-reliance on high probability labels.
• Backwards compatible, in line with semantic versioning.

Version 1 - First Release

The first finished product of Whichpet, ready for use as an external library.

It includes the whichpet.js and the API described in the README.md, as well as a working example of using the whichpet.js in the form of a single html page and a .json file that includes the initial data set.