Sentiment aware interactive Chatbot AI using multi agent processing model

Abstract

Understanding user sentiment has become more important for organizations and consumers due to the rapid growth of social media platforms such as marketplaces, platforms for connecting brands and consumers, and public discussion platforms. Emotions that are based on aspects, nuanced within context, and multifaceted often require complex sentiment analysis algorithms to interpret properly. Furthermore, these systems do not provide real-time information to help companies make better decisions and enhance consumer satisfaction. To tackle these challenges, a novel Interactive Chatbot artificial intelligence (IChat-AI) approach has been proposed in this paper for sentiment-aware chatbot interaction. The word to vector (W2V), term frequency-inverse document frequency (TF-IDF), and bag of words (BoW) are utilized to effectively extract essential features. The deep Kronecker neural network (DKNN) is utilized to predict and classify the emotions into five classes, such as sad, happy, neutral, angry, and fearful. Python has been used to simulate the suggested model. The efficacy of the suggested system is examined employing parameters including recall, execution time, F1-score, complexity, precision, scalability, accuracy, and response time. The developed IChat-AI strategy performs better regarding accuracy than the existing methods, including RoBERTa, TLSA, and multimodal transformers fusion for desire, emotion, and SA (MMTF-DES) approaches, by 5.33%, 4.73%, and 14.39%.