CAM-Bot: A Context-Aware Multimodal Chatbot for Enhanced E-commrce Customer Support

Abstract

This paper introduces CAM-Bot, a novel Context-Aware Multimodal Chatbot designed to revolutionize customer support across diverse digital business domains. Leveraging advanced natural language processing, computer vision, and context modeling techniques, CAM-Bot demonstrates significant improvements over existing systems in handling complex, multi-turn dialogues and multimodal interactions. Our experiments, conducted on subsets of MultiWOZ 2.1, Amazon Review Data, and VQA v2.0, show that CAM-Bot outperforms state-of-the-art baselines across key metrics. Notable improvements include a 3.5% increase in BLEU-4 score (from 15.01 to 15.53) for task-oriented dialogues, a 2.6% gain in Success Rate (from 70.1% to 71.9%) for task completion, a 2.6% improvement in NDCG@10 (from 0.4912 to 0.5038) for e-commerce recommendations, and a 0.9% accuracy boost (from 66.14% to 66.72%) in visual question answering. Through our analysis, we identify the crucial role of context aggregation and highlight areas for further enhancement. We explore practical applications of CAM-Bot in various sectors, including e-commerce, financial services, and technical support, demonstrating its versatility. This research not only showcases the potential of integrating context-awareness and multimodal processing in chatbots but also provides valuable insights into the future of AI-driven customer support in digital business environments. CAM-Bot’s adaptable architecture paves the way for more intelligent, responsive, and personalized customer in-teractions across a wide range of digital platforms and services.