Neuro-Symbolic AI for News Summarization: A Hybrid Ontology and Deep Learning Approach

Abstract

In an environment where the sheer volume of news articles makes it challenging for readers to quickly access relevant information, traditional text summarization techniques often fail to maintain contextual richness and factual accuracy. Ontology-based prompt tweaking and abstractive summarizing techniques are combined in this study's hybrid approach to provide news summaries that are accurate, succinct, and contextually relevant. By embedding domain- specific knowledge through ontologies into the summarization process, the study addresses the shortcomings of conventional summarization models and enhances the relevance and clarity of the generated outputs. Motivated by the increasing demand for efficient information retrieval, the research develops a framework that automates summarization through structured knowledge representation and iterative prompt refinement. Using a dataset of sports news articles alongside a corresponding sports ontology, the approach demonstrates practical improvements in summary quality, supporting better decision-making and knowledge dissemination. The proposed model aims to deliver domain-aware, fluent summaries that improve user experience in navigating overwhelming volumes of news content.