v4.2.0

v4.2.0

• Library:
  • Refactoring: moved Serializations namespace to NW.Shared.Serialization.

v4.1.0

v4.1.0

• Library:
  • Refactoring: moved Files and Validation to NW.Shared.Files and NW.Shared.Validation.

v4.0.0

v4.0.0 - BREAKING CHANGES

• Library:
  • Feature: "Bag Pattern" adopted into the codebase (*Settings => SettingBag, *Components => ComponentBag, ...).
  • Framework Freeze: migrated projects to .NET 8.0 and NUnit 4.x; updated packages accordingly, implemented new NUnit assertion model.
• Documentation:
  • Feature: updated to v4.0.0.

v3.7.0

v3.7.0

• Library:
  • Performance: ClassifyMany() now performs 75% faster thanks to parallelism.
• Documentation:
  • Feature: added Documentation-NW.NGramTextClassificationScripts.md.
• Scripts:
  • Feature: added ngramtchelper.ps1.

v3.6.0

v3.6.0

• Library:
  • Feature: added support for the "CleanLabeledExamples" functionality.
  • Feature: added TextSnippet property to TextClassifierResult.
  • Feature: added support for the "disableindexserialization" flag for the SaveSession method.
  • Performance: the tokenization process of the provided labeled examples now happens only once, despite of the number of text snippets.
• Client:
  • Feature: added "--cleanlabeledexamples" option to the "classify" sub-command
  • Feature: added "--disableindexserialization" option to the "classify" sub-command
• Documentation:
  • Feature: updated to v3.6.0.
• File Formats:
  • Feature: added TextSnippet field to each result object to Session Json.
  • Feature: added a Session Json's variant with disabled index serialization.

v3.5.0

v3.5.0 - BREAKING CHANGES

• Library:
  • Feature: added support for the classification of a collection of snippets of texts.
  • Feature: added support for Minimum Accuracy* fields in TextClassifierSettings.
  • Feature: added support for "Files" namespace.
  • Feature: added support for "Filenames" namespace.
  • Feature: added the possibility to load and save labeled examples from/as JSON file.
  • Feature: added the possibility to load and save text snippets from/as JSON file.
  • Feature: added the possibility to load a NGramTokenizerRuleSet from a JSON file.
  • Feature: added the possibility to save a TextClassifierSession as JSON file.
• Client:
  • Feature: added "classify" sub-command and its options.
• Documentation:
  • Feature: updated to v3.5.0.

v3.0.0

v3.0.0 - BREAKING CHANGES

• Decoupled LabeledExample class from the tokenization process; TokenizedExample class added.
• Removed all the Invoke() statements to make the code is more compact.
• Split the monolithic ObjectMother class to namespace-based ObjectMother classes to improve maintainability.
• Split the monolithic MessageCollection class to namespace-based MessageCollection classes to improve maintainability.
• TextClassifier, added LogAsciiBanner() method.
• Added a preliminary client structure for NW.NGramTextClassificationClient.