Build Training set for main.

Using the jupyter notebook algo I wrote. Alter that to include other possible typos and what not.

1. Then build a connector to write our training data to the table.

2. Then another connector to pull that data for training a model in Jupyter.

Train model by active labeling.


multithreading to read at intervals from the same folder as our model and pull results to write to db.


Consumer that would connect to a broker and consume new records for deduping.