高质量数据集产品的形态和生产流程研究OA
Research on the form and production process of high-quality dataset products
高质量数据集决定了人工智能模型的训练效果.高质量数据集缺乏统一标准形态和质量可控的流程化生产方法,导致其供给不足、流通不畅,已成为制约人工智能发展应用的因素之一.本文从数据产品的角度,提出高质量数据集产品的五元组形态,以全链路技术能力为支撑,设计高质量数据集产品的生产流程,提出面向产品需求的全生产流程质量管控方法,为高质量数据集产品的大规模生产、流通提供理论基础和可行方案.
High-quality datasets determine the training performance of artificial intelligence models.The lack of a unified standard form and a quality-controllable,process-based production method for high-quality datasets has led to their insufficient supply and inefficient circulation,which has become a bottleneck restricting the development and application of artificial intelligence.From the perspective of data products,this paper proposes a five-tuple form for high-quality dataset products.Supported by full-link technical capabilities,we design a production process for such products and propose a product-oriented quality control method that covers the entire production chain.This work provides a theoretical basis and feasible solutions for the large-scale production and circulation of high-quality dataset products.
杨琳;朱扬勇
上海市大数据中心,上海 200003||华东师范大学数据科学与工程学院,上海 200062上海数据研究院有限公司,上海 200120
信息技术与安全科学
高质量数据集数据产品形态生产流程数据质量
high-quality datasetdata productformproduction processdata quality
《大数据》 2026 (2)
64-74,11
评论