Skip to main content

Standard Procedures (Natural Language Processing Chatbot) Overview

This Standard Procedure will show you a step-by-step guideline on building a chatbot with natural language processing capability. It will also highlight the tips that you should pay attention to during your chatbot building journey.

Level One - Apply NLP to your Chatbot

Apart from text or payload trigger, you can actually utilise NLP (Natural Language Processing) triggers. This means that WOZTELL can pass the user input to NLP engines for analysis, receive the correct intent and then give out an appropriate answer.

Find out how to add capability to your chatbot here.

Level Two - Set Up an NLP Fallback Tree

Your Chatbot is not superman. There are learning curves and it takes time for your Chatbot to absorb knowledge and new data. However, your chatbot should try its best to cope with any user input it being thrown at.

One way to significantly improve the chatbot capability of answering a human question is to add an NLP fallback mechanism to your conversation flow.

Find out how to add an NLP fallback mechanism here.

Level Three - NLP Fallback to Other Languages (Coming Soon)

Sometimes the reason behind why your NLP chatbot cannot detect a user input is not because you didn't train your intent good enough, but it is because the user input is in another language and your NLP agent can only recognize 1 language at a time.

Good thing is that you can actually utilise multiple NLP training agents on WOZTELL.

Find out how to allow your NLP chabot to detect multiple languages here.