Comparative Analysis of Classical and Neural Networks based ChatBot’s Techniques

Abstract

Conversational agents like Alexa from Amazon, Siri from Apple, Assistant from Google, and Cortana from Microsoft demonstrate extraordinary research and potential in conversational agents. A conversational agent, chatter-bot, or chatbot is a piece of computer software supposed to communicate at a level of intelligence comparable to a person's. Chatbots are designed for various purposes, such as task-oriented helpers and creators of open-ended discourse. Numerous approaches have been studied, from primitive types of hard-coded response generators to contemporary ways of constructing artificial intelligence. These are classified as rule-based or neural network-based systems. Unlike the rule-based technique, which is based on pre-defined templates and responses, the neural network approach is based on deep learning models. Rule-based communication is optimal for more straightforward, task-oriented conversations. Open-domain conversational modeling is a more complicated topic that depends heavily on neural network techniques. This article begins with an overview of chatbots before diving into the specifics of a variety of traditional, rule-based, and neural network-based methods. A table summarising previous field research closes the survey. It looks at the most recent and vital research on the subject, the evaluation instruments used areas for improvement, and the applicability of the proposed methods.