Retrieval-Augmented AI Chatbot for Real-Time News Summarization and Fact Verification

Abstract

Conventional news bot application relies heavily on great datasets and deep learning models for computing. This paper presents a new approach for news retrieval and summarization using AI-powered Sparse Predictive Hierarchies. In contrast to deep learning- based methods, SPH allows for incremental learning, with a low footprint that makes it lightweight, adaptive, and suitable for dynamic environments. The systematic chatbot exploits the abilities of Retrieval-Augmented Generation to search for news articles most relevant to the input query and generates context-aware responses. Neuro-symbolic reasoning supplements the ability of the chatbot to process news with better interpretability and decision-making. This approach enhances adaptability and reduces latency and computational costs, making it ideal for deployment in resource-constrained devices and real-time applications. Experimental results demonstrate that the chatbot is faster and more contextually accurate than its conventional deep learning counterparts.