SL_NLP_PassiveActive1

Run notebooks

Run the notebooks in the exact order. The output files from each notebook is the input for the next notebook.

0. Place an 'input.csv' file in the root folder. It should have 2 required columns which are "prompt" and "response".
1. Run spacy_classifiers.ipynb. This will split the responses into clauses in the file voice_classified.csv
2. Run abstraction_scores.ipynb. This will split the responses into clauses in the file abstraction_scored.csv
3. Run readability_scorer.ipynb. This will split the responses into clauses in the file readability_scored.csv
4. Finally run final_output_nb.ipynb. This will produce two files output.csv and debug.csv. Output.csv is the minimal output which contains the split clauses, the final score and final voice based on maximum internal scores. Debug.csv contains a bit more granular details and scores of each internal terms.

Run locally

Steps to set up an internal tool are:

1. Clone this project. Move to the project folder.
2. Run pip install -r requirements.txt
3. Run python -c "import nltk; nltk.download('wordnet'); nltk.download('stopwords'); nltk.download('punkt');"
4. Run python -m spacy download <model-type>. Model type can be en_core_web_lg, en_core_web_md or en_core_web_sm.
5. Run python voice_identifier.py --help to get started.