Chatbot using RASA NLU

A chatbot has many applications ranging from customer service in sales and marketing department to product recommendations in E commerce websites. 

There are many types of chatbots, it may be a rule based chatbot where there is a sequence of questions and answers aligned together one leading to the other. This can be build using any hard coded programming language. But is this chatbot user friendly and dynamic? The answer is No, though it serves the purpose but people are more inclined towards building a dynamic and user friendly chatbot which is more instinctive in its responses and can handle queries which are not a part of the training data. 

An easy way to build such a chatbot is using RASA NLU. RASA is an open-source framework developed in a very user-friendly way to build chatbots. 

It is made up of three parts-Intent, Stories, Actions/Reponses. Intent is a user defined user defined category, and any query which is asked to the chatbot falls under one of the intent categories. For every intent there is a response or an action which will direct the user to the next step as per his query. 

The interesting part of RASA chatbot is the way the mapping is done. For every intent a response is mapped. Different queries are provided as a training data for the intent. 

Eg- Intent-Welcome- queries like hi, hello, good morning, good evening, need help, etc. fall under intent welcome. To which there is a response saying “Hi Welcome! How do I Help You? “ 

The third part of the chatbot is the “Stories”. Stories are a flow of how one intent will lead to next intent and so on with every intent assigned a response. So, stories are a hierarchy of queries. 

All these three parts can be altered and the model can be trained to generate a very efficient chatbot. The responses can be made customized using an action.py script as per what the user wants. 

The chatbot has a fallback action, can be assigned forms, has a database connectivity, and hence covers all the elements needed to create an easy but efficient chatbot. 

For more details refer the RASA Documentation- https://rasa.com/ 

Leave a Comment

Your email address will not be published. Required fields are marked *