This paper explores the use of large language models (LLMs) as an alternative to traditional topic modeling approaches for uncovering underlying topics within extensive text corpora. The authors investigate the potential of LLMs by introducing a framework that prompts LLMs to generate topics from given documents and establish evaluation protocols to assess their clustering efficacy. The findings indicate that LLMs, with appropriate prompts, can effectively generate relevant topic titles and adhere to human guidelines for refining and merging topics. Through experiments and evaluations, the paper summarizes the advantages and constraints of using LLMs in topic extraction, highlighting their ability to handle dynamic datasets and provide interpretable results. The study also proposes evaluation metrics to measure the quality of topics generated by LLMs, demonstrating their suitability for both labeled and un labeled datasets. A case study on temporal analysis of COVID-19 vaccine hesitancy further showcases the practical applications of LLMs in real-world scenarios. Overall, the paper provides a comprehensive exploration of LLMs as a viable and adaptable method for topic extraction, offering a fresh perspective compared to traditional topic modeling techniques.This paper explores the use of large language models (LLMs) as an alternative to traditional topic modeling approaches for uncovering underlying topics within extensive text corpora. The authors investigate the potential of LLMs by introducing a framework that prompts LLMs to generate topics from given documents and establish evaluation protocols to assess their clustering efficacy. The findings indicate that LLMs, with appropriate prompts, can effectively generate relevant topic titles and adhere to human guidelines for refining and merging topics. Through experiments and evaluations, the paper summarizes the advantages and constraints of using LLMs in topic extraction, highlighting their ability to handle dynamic datasets and provide interpretable results. The study also proposes evaluation metrics to measure the quality of topics generated by LLMs, demonstrating their suitability for both labeled and un labeled datasets. A case study on temporal analysis of COVID-19 vaccine hesitancy further showcases the practical applications of LLMs in real-world scenarios. Overall, the paper provides a comprehensive exploration of LLMs as a viable and adaptable method for topic extraction, offering a fresh perspective compared to traditional topic modeling techniques.