Memory-Enhanced Conversational AI: A Generative Approach for Context-Aware and Personalized Chatbots

Abstract

This research addresses the limitations of conventional conversational chatbots, which often provide generic responses, resulting in a lack of engaging interactions. The study introduces an advanced memory storage and retrieval system to enhance the chatbot's ability to remember past conversations, focusing on context awareness and personalization. The goal is to create a more seamless and dynamic conversational experience, alleviating user frustrations and elevating overall satisfaction. The proposed solution extends beyond immediate concerns, contributing to improved natural language processing (NLP) skills and fostering intelligent, adaptable, and user-centric conversational AI. The methodology involves data collection from a diverse dataset, employing a distilled GPT-2 tokenizer for text preprocessing, and implementing a generative-based model for context-rich responses. Validation metrics encompass fluency, user satisfaction, memory recall, perplexity, diversity, and consistency. The research concludes with successful results, demonstrating the effectiveness of the chatbot in addressing user concerns and contributing to the advancement of conversational AI.