Online and social media analysis has been revolutionized in recent years by the development of Large Language Models (LLMs) and their ability to process unstructured text. While prior research has demonstrated their strong performance across a wide range of tasks, the integration of semantic tools such as Knowledge Graphs with LLMs represents a more recent and impactful advancement. This work proposes an approach that combines the computational capabilities of LLMs with Knowledge Graphs and Rough Set Theory in a unified operational pipeline, designed to enhance the semantic understanding of online debate dynamics. The pipeline is validated through a case study of Reddit conversations related to the Israeli conflict. The results provide enhanced insight into these dynamics, supporting decision-makers and analysts in their interpretation.