<p>Meetings involve complex multi-party discussions where key insights, decisions, and consensus emerge through interactive deliberation. Automatically summarizing such interactions requires capturing not only salient content but also the underlying discourse structure that reflects agreement among participants. However, most existing summarization approaches process meeting transcripts as linear sequences, failing to account for the dialogic and collaborative nature of real-world discussions. In this study, we introduce a novel abstractive summarization framework that emphasizes consensus by modeling meetings as structured interactions centered on points of agreement. Our approach identifies the flow of topical discourse and highlights segments where consensus is reached, enabling the generation of summaries that more accurately reflect the outcomes and resolutions of the conversation. Through comprehensive experiments and analyses, we demonstrate that the proposed framework substantially improves the factual alignment and informativeness of generated summaries, achieving superior performance over a range of competitive baselines. Furthermore, qualitative evaluation shows that the summaries produced by our method are more faithful to the actual decisions and shared conclusions articulated during meetings, underscoring the value of incorporating consensus-awareness into summarization systems.</p>