Build a demo application

[ ] Build a number of architectures incorporating data processing, lemmatizers encoders and models to map intents
[ ] handle word formatting and missing words, remove punctuation for lemmatization
[ ] Build models to identify and catalogue objects in the text (eg. orders etc.) Could use regex.
[ ] Process and categorize email history into a categorised training set
[ ] May need to consider data augmentation
[ ] Build dialogue trees for each intent
[ ] Train the models using them to 'vote' on the answer and learn simultaneously as much as possible.
[ ] Track model performance
[ ] Handle the uncertainty problem - stay quiet when uncertain
[ ] Handle continuous, ongoing training
[ ] Automatically download new email input
[ ] Track email threads - don't consider new unless a new thread
[ ] Switchable feature to auto-reply
[ ] Package with PyOxidizer
[ ] Deploy as an aid during testing
[ ] Deploy without restriction
[ ] May need to consider data augmentation (https://machinelearningmastery.com/how-to-configure-image-data-augmentation-when-training-deep-learning-neural-networks/)

Colin-Codes / IntentClassifier-ML-Project