E-mail classification
Problem
Emails are the most convenient way of written communication these days, but they can be challenging for the business when it requires users, especially receivers, to spend an extra amount of time shifting through the mass of emails and the details that they contain.
Solution
An automated approach that can reduce the time to classify emails, learn the details associated with it and then ultimately communicate back with a set of appropriate information.
Results
An NLP machine learning model was created, classifying incoming emails and attributing them to a response template. The pre-conditions were that of having only 25% of emails classified, which left solid room for improvement. Ruled Based and Machine Learning combination was confirmed to be the best approach, exhibiting over 70% accuracy.
