Intelligent Chatbot Using GNMT, SEQ-2-SEQ Techniques

Abstract

Dialogue generation and intelligent chatbot development is a fairly new avenue in the field of artificial intelligence and machine learning. Chatbot development is a popular field in natural language processing. Nowadays a large amount of research projects are currently underway to make efficient and reliable chatbots using machine learning, artificial intelligence and natural language processing. Chatbots or dialogue generation agents are mainly used by businesses like online stores or startups, financial institutions like banks, credit card companies and govt organisations. There are many different types of chatbots available differing in frameworks such as code based or interface based. But these chatbots lack in genuine dialogue generation and seem rather bland and unintuitive. Some of the popular intelligent chatbots or you can say personal assistants include google assistant, Siri by apple and cortana by microsoft. The functioning of these chatbots is constrained and limited and they are not aimed to hold real world conversations. In this project endeavour we have tried to develop an intelligent chatbot using state of the art GNMT(Google neural machine Translation) model which is built upon seq-to-seq modelling having encoder-decoder architecture. This encoder decoder discussed here uses recurrent neural networks along with bi directional LSTM(Long short term memory) which is one of the hot areas.