Analysis of Live Emotionally Intelligent Bot

Abstract

Emotion is a human way of expressing feelings. Emotions tell how well the social interactions of the user went in the recent past. This can prove to be valuable information to improve and enhance user experience in chatbots. This information will make bots not just intelligent but emotionally intelligent by understanding the user’s social life. The information can help in providing more relevant answers to the user. The user messages in chatbots do not provide any information about the user’s state and emotions. There is a need for some other data source that could provide valuable information about the user’s emotions.
In this project, we have developed an emotionally-intelligent bot application based on a convolution model. The goal is to improve the performance of bots by using the facial emotions of the user. The convolution model is trained using TensorFlow on 48x48 pixel gray-scale images of faces. The assessment of the performance of bot is also discussed. The outcomes conclude a significant increase in user satisfaction.