Mona Lisa chatbot

Abstract

Conversational chatbots are becoming more and more pervasive in our society and culture. The seamless communication between man and machine is always at the forefront of Artificial Intelligence. This project is particularly focused on chatbots that can exhibit elements of a persona. The motivation behind this goal is that conversation between man and machine should be as natural as possible. The artifact of this project was designed with the intention of implementing an element of a persona, namely emotion. This was done by creating a chatbot designed to determine its response based on the emotion that is attributed to the user's input. The system uses Machine Learning techniques to achieve this goal. The end result is a proof-of-concept system that can converse according to the state of emotion. The intention of this work is to emphasise the concept and practices that already go into developing a persona, whilst also putting emphasis on the development of a persona's emotional intelligence. The results of the artifact showed clear success in the development of an emotion detection system, and somewhat good results in the response retrieval system, that could be bettered with a more vast dataset.