Seconds of Meeting Using NLP

This paper presents an innovative paradigm that would reshape the consumption and retrospective analysis of meetings by implementing the Natural Language Processing (NLP) and other advanced machine learning paradigms. The system proposed is called Seconds of Meeting, and it represents a smart meeting summarization system, which abridges the lengthy discussions in concise and impactful summaries. In the case where some meetings last more than a few hours, practitioners that cannot be present during them or have limited time to do so can have on the spot a one- page summary that will capture the main arguments, decisions, and action plans. Using state of the art applications like Pegasus, LongT5 and Whisper, the system translates verbal dialogue into accurate textual information, which is then processed into a summarized form, which in turn, can be understood in a minimum of 59 seconds. The empirical evaluation shows that the system achieves a total transcription and summarization fidelity of approximately 80, thus, providing reliable performance in a variety of meeting settings. At the end of the session, the summary is automatically created as a PDF document, which allows free documentation and availability. Such combination of NLP, Automatic Meeting Logging (AML), and Large Language Models (LLMs) promotes effective information storage, temporal reduction, and boosts productivity at both the business and academic level.