Merge branch 'main' of https://github.com/xehu/team-process-map into …

…main

feature_engine/README_featurizer.md

@@ -4,22 +4,10 @@ The featurizer takes text conversations and transforms them into the following r
 
 - A table of `chat-level features`, which generates a unique conversation feature for each chat (aka utterance, or message);
 - A table of `conversation-level features`, which generates aggregations of features within each chat at the conversation level.
+- A table of `user-level features`, which generates aggregations of features for each user, or speaker, in a conversation.
 
 To set up and run the featurizer from scratch, you should do the following.
 
-# Run separate iPython Scripts
-Some features are computationally inefficient to run every time, so the featurizer performs some processing upfront. Before getting started, you should separately run the following iPython notebooks.
-
-## Run ONCE (regardless of number of datasets)
-- `features/preprocessing/preprocess_lexicons.ipynb` --> generates `features/lexicons_dict.pkl`
-
-## Run once _per dataset_ (generates dataset-specific pre-processing / embeddings)
-The following needs to be run upon initializing the directory:
-- `features/preprocessing/process_sent_vectors.ipynb` --> generates `embeddings/*`
-
-The following does not have to be run upon initialization (the outputs are already saved); however, as new datasets are added, this script needs to be re-run for each new dataset.
-- `features/preprocessing/positivity_bert_analysis.ipynb` --> generates `sentiment_bert/*`
-
 # Run the main featurizer [Do this every time you want to refresh/generate new features.]
-
-In the terminal, run `python3 featurize.py`.
+- Declare a new FeatureBuilder object inside `featurize.py`.
+- In the terminal, run `python3 featurize.py`.