Note: I'm new to Machine Learning and NLP. This is my first project in this field.

I'm using NaturalNode/natural (https://github.com/NaturalNode/natural) to build a chat bot to help my users with common questions.

I've implemented a BayesClassifer (http://naturalnode.github.io/natural/bayesian_classifier.html) in which I have supplied a number of questions and answers to train it on.

Example of data I am training it on using the 'addDocument' call like the example shown on the page above:

  • I've lost my password
  • What is my password
  • what are my login details
  • I can't remember my password
  • how do I reset my password
  • how do I change my password
  • when will I get paid
  • Where can I find my earnings
  • Why havne't I been paid?
  • Can I get paid today?
  • How do I setup my domain?
  • What is my domain?
  • Can you give me a free domain?
  • why is my domain deleted?
  • I need to change my domain

I'm only want to return an answer if the confidence in the accuracy is high. However, when I run a classification, even if it's an exact match, the confidence values are really low (sub .4). I'm wondering if there are more steps I should take to improve the confidence like removing noise words from the training values or the question.

Are there further steps I can take to improve the matching confidence for this process?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.