- Model : CNN model with pre-trained embeddings
- Model : Baseline models (gender word and toxicity)
- Model : CNN model with emebedding layer
- Feature : BERT document embeddings preprocessing and feature extraction
- Experiments : Save the classification performance and the parameters of the model.
- Experiments : Balance classes for training and testing (with downsampling)
- Run pipeline
- Test data domain preparation
- Feature selection for ngram and type dependency feature sets
- The hyper-parameters tuning with GridSearchCV
- Logistic Regression model
- Training data domain preparation
- Run pipeline
- Type dependency preprocessing and feature extraction
- Training data domain preparation
- Simplified notebooks.
0.0.4 - 2020-09-14
- CHANGELOG.md file
- README.md file
0.0.3 - 2020-09-11
- Attributes added in setup.py file to run the project on GESIS notebooks.
0.0.2 - 2020-09-10
- Create sklearn pipeline with transformer and featureunion.
- setup.py file
- requirements.txt file
- .gitignore file
- Abstract classes for preprocessing and feature extraction steps.
- Neutral and Compound scores use for sentiment feature extraction.
0.0.1 - 2020-09-02
- Preprocessing step for Sentiment and Ngram features.
- Feature Extraction step for Sentiment and Ngram features.