All the details in https://www.kaggle.com/c/tweet-sentiment-extraction/data (competition: https://www.kaggle.com/c/tweet-sentiment-extraction) also apply to this new dataset. The profiles text data has been generated using the data preparatory notebook https://www.kaggle.com/neomatrix369/fastchai-tweet-sentiment-extraction-data-prep with the help of the [NLP Profiler](https://github.com/neomatrix369/nlp_profiler) library. In addition to the standard `train.csv`, `test.csv` and `submission.csv` files, we have these additional files generated by profiling the respective text fields in `train.csv` and `test.csv` (see https://www.kaggle.com/c/tweet-sentiment-extraction/data for definitions of the respective columns). **Generated from `train.csv`** *from the column `text`* - `profiled_train_text_granular_features.csv` - `profiled_train_text_grammar_check.csv` - `profiled_train_text_sentiment.csv` - `profiled_train_text_spelling_check.csv` *from the column `selected_text`* - `profiled_train_selected_text_granular_features.csv` - `profiled_train_selected_text_grammar_check.csv` - `profiled_train_selected_text_sentiment.csv` - `profiled_train_selected_text_spelling_check.csv` **Generated from `test.csv`** _from the column `text`_ - `profiled_test_text_granular_features.csv` - `profiled_test_text_grammar_check.csv` - `profiled_test_text_sentiment.csv` - `profiled_test_text_spelling_check.csv` Each of the above datasets (`.csv` files) do not contain the respective `text` or `selected_text` columns due to memory allocation issues when processing in the notebook.