CUTTING-EDGE CHAT-BOT SENTIMENT ANALYSIS MODEL USING EXTENDED SCALE FOR POLARITY DETECTION
Dr. M. Devi Sri Nandhini1, Dr. G. Pradeep2
1Assistant Professor-III, 2Associate Professor
School of Computing, SASTRA Deemed University, Thirumalaisamudram, Tanjore, Tamil Nadu, India
ABSTRACT: Chatbots can comprehend human moods and emotions thanks to sentiment analysis. People are emotional beings. When your consumers engage with a chatbot for customer support, they anticipate that the bot will be able to discern their emotions, tone, and mood and respond accordingly. Assessing consumers’ interactions with chatbots is one of the fascinating domains in which sentiment analysis finds use. AI chatbots have been developed for customer support in an effort to improve customer service and lower corporate costs. Chatbot sentiment analysis is the process of identifying a user’s words and voice when they interact with a chatbot, regardless of their emotional state—happy, sad, afraid, or angry. Customers aren’t guaranteed to forget or disregard a negative encounter just because they dealt with a chatbot. This clearly indicates the importance of carrying out chatbot sentiment analysis in order to better understand the emotions of customers and thereby offer a tailored response. The intricacies of conducting sentiment analysis on chatbots are covered in this study. It outlines the process for conducting sentiment analysis on chatbots and demonstrates how to automatically identify sentiment in talks between chatbots and customers. Furthermore, the suggested approach investigates the analysis and classification of text sentiment as well as the use of chatbot sentiment analysis by businesses to automatically categorize customer comments as positive, negative, or neutral, helping them to accurately grasp consumer sentiment. The proposed work employs the extended sentiment scale (nonary scale) developed in ERBA-DSL model of our previous work to detect the polarity of the sentiment in chatbot-customer interaction and achieves a reasonably good F1-score of 84%.
KEYWORDS: Chatbot Sentiment Analysis, Chatbot-Customer Interaction, Sentiment-Aware Chatbot Response Generation.