Exploration of High-Quality Dataset Development and Application Practices in Basic Disciplines
|更新时间:2026-08-19
|
Exploration of High-Quality Dataset Development and Application Practices in Basic Disciplines
Big Data Research(2026)
作者机构:
1.中国科学院计算机网络信息中心,北京 100083
2.国家基础学科公共科学数据中心,北京 100083
作者简介:
基金信息:
National Key R&D Program of China“Research and demonstration on key technologies of collaborative services of multinational scientific data”(2025YFE0211200)
Gao Yuwei, Zhu Yanhua, Hu Lianglin, et al. Exploration of High-Quality Dataset Development and Application Practices in Basic Disciplines[J/OL]. Big Data Research, 2026.
DOI:
Gao Yuwei, Zhu Yanhua, Hu Lianglin, et al. Exploration of High-Quality Dataset Development and Application Practices in Basic Disciplines[J/OL]. Big Data Research, 2026.DOI: 10.11959/j.issn.2096-0271.BDR26217.
Exploration of High-Quality Dataset Development and Application Practices in Basic Disciplines
在AI for Science驱动全球科研范式变革的背景下,高质量科学数据集已成为支撑基础学科创新的核心生产要素,其建设质量直接关系到我国基础研究核心竞争力。针对当前科学数据资源分散、标准不统一、质量参差不齐、成果转化不畅等行业痛点,国家基础学科公共科学数据中心立足多学科发展需求,探索形成多维度协同的高质量数据集建设体系。本文结合中心实践成果,系统阐释基础科学高质量数据集的核心内涵,提出培育权威数据库、遴选AI4S数据集、形成用户高评价数据集、建设数据论文预印本平台、升级智能体服务体系等五大实践路径,剖析数据应用价值与转化机制,可为我国基础科学数据集规范化建设、开放共享及AI-ready成果培育提供理论参考与实践借鉴。
Abstract
Driven by the global research paradigm transformation of AI for Science
high-quality scientific datasets have become core productive resources for the innovative development of basic disciplines
fundamentally determining the competitiveness of China’s basic research. To address the prevalent issues of fragmented data resources
inconsistent standards
uneven data quality
and inefficient transformation of research outcomes
the National Basic Science Data Center has constructed a systematic
multidimensional framework for high-quality dataset development. Based on the center’s practical experience
this paper expounds the core connotation and construction criteria of basic science datasets and summarizes five key practices: the development of authoritative databases
the construction of AI for Science-oriented dedicated datasets
the integration of user‑feedback‑optimized high-quality datasets
the establishment of a scientific data preprint platform
and the deployment of an AI-based intelligent service system. This study further analyzes the application value and transformation mechanism of scientific data. The findings provide theoretical support and practical references for the standardized development
open sharing
and AI-ready output of basic science datasets in China.