Smart Chatbot System with Hybrid AI for Advanced Sentiment Analysis

Chatbots facilitate digital communication for human users through the utilization of natural language processing (NLP) driven by artificial intelligence (AI). These chatbots have undergone prior training to comprehend user inquiries and promptly deliver responses employing natural language processing. User inputs directed towards chatbots may manifest in various formats, encompassing voice, text, and sentiment. In spite of extensive research endeavors, extant chatbots frequently grapple with issues such as inaccurately discerning user needs, providing irrelevant responses, and grappling with challenges in voice-based analysis and deciphering user intentions, particularly in prolonged dialogues. Consequently, familiarity with the context and execution of sentiment analysis hold pivotal significance. This study advocates for the enhancement of chatbot responses through the employment of Bidirectional Encoder Representations from Transformers (BERT) for contextual comprehension, Support Vector Machines (SVM) integrated with Fuzzy Logic for sentiment analysis, and the Text-To-Text Transfer Transformer (T5) for tasks involving sequence-to-sequence functionality. This approach aims to enhance the precision of human interactions by formulating chatbot models utilizing AI-driven sentiment analysis. Chatbots are grounded in two realms of AI machine learning and natural language processing, with applications spanning across diverse domains such as digital marketing, education, and virtual communities. The proposed BERT-SVM-Fuzzy Logic-T5 model exhibited an accuracy level of 93% through the utilization of the T5 technique for sequence modeling, and 92% in its absence.