The work aims to analyze a large number of online reviews to uncover market requirements and use big language models to generate more accurate and comprehensive design requirements, so as to improve the ability of the contract-issuing party in crowdsourced design and accurately identify and describe design requirements. By scraping online review texts from e-commerce platforms, the system identified and analyzed the subjects of the reviews, categorized them using spectral clustering, calculated user satisfaction, and generated requirements through large language model dialogues. This method systematically analyzed user feedback from online reviews, identified and clustered the subjects of the reviews, and then identified and generated user requirements. Experiments demonstrated that this method effectively generated requirements of crowdsourced design. This research successfully integrates online review-based requirement mining and large language models to identify and generate design requirements, achieving precise targeting and efficient generation of design requirements and providing a new approach for crowdsourced design.