v1.3.1

Bugfixes

• Lexicon filtering used in emotion and psycholinguistic features now takes into account all lemmas, not unique ones; addressing #17

Performance improvements

• Filtering lexicons rewritten to polars join operations instead of element mapping, benefiting from polars ' internal optimization.

Changes

• Added pointer to READMEs for users to fix potential wn incompatible database schema errors.

v1.3.0

Version 1.3.0

New Features

• Maximum (max), Minimum (min), and Standard Deviation (std) of psycholinguistic and emotion features

Extended Multilingual Support

• Added psycholinguistic norms for additional languages:
  • Concreteness:
    • Spanish (es)
    • Polish (pl)
  • Age of Acquisition (AoA):
    • Spanish (es)
    • Dutch (nl)
    • Polish (pl)

Bug Fixes

• Included a tutorial on how to download the NRC lexicons in the documentation and READMEs to address licensing restrictions.
• Fixed the multilingual support table to correctly reflect the available psycholinguistic norms for each language.
• Added an indication for feature group extraction, if the respective group is not found, to improve debugging for users.

Optimizations and Improvements

• Refactoring lexicon/norm-based features to allow for easier maintenance and extension: Added generic feature implementations to reduce code duplication and apply improvements across multiple features automatically.
• Clearer custom extraction documentation.

v1.2.0

New

• Integrate psycholinguistic features for German, Italian, and French:
  • concreteness (de, it, fr)
  • age of acquisition (de, it)
  • sensorimotor (it, fr is postponed)
• Warnings for empty text instances and if a feature contains NaNs

Bugfixes

• Initializing Extractor with a dataframe now does not do unwanted changes on the original dataframe anymore
• Fix wrong backend mentioned in the configuration tutorial

What's Changed

• Add tests + token normalization functionality by @mmmaurer in #2
• Create postbuild by @ArthurMuehl in #3
• Update README_MH.md by @ArthurMuehl in #4
• Bundle models for mybinder.org by @rgaiacs in #6
• Install Python package as editable for mybinder.org by @rgaiacs in #5
• Minor spell correction by @rgaiacs in #7
• v1.2.0 by @mmmaurer in #8