摘要
随着全球各科学领域大科学装置的出现,科学发现进入了大数据时代。科学发现无法完全依赖于专家经验从海量数据中发现稀有科学事件,大量历史数据无法有效利用,同时愈发突出实时性和高精度,科学事件的模式具有稀有性,通用的算法并不适用于科学领域,由此科学数据智能发现问题应运而生。科学数据智能发现旨在使用数据智能的方法加速科学事件的发现。然而,科学数据智能发现缺少整体框架设计,具体表现为缺乏科学数据的一体化分析体系和异构科学数据高效知识融合机制,并且海量历史数据长期存储及挖掘低效。本文从数据管理的角度提出科学数据智能发现与管理框架和相关挑战,以期推动科学发现的进步。
The large-scale scientific infrastructure has been accelerating all fields of science into Big Data Era.Although many interesting scientific events are contained in such a huge amount of data,it brings many a lot of trouble to scientists.Scientists can no longer rely on their experience to discover rare scientific events from massive data as they did before.The data intelligence technology is one of important topics to discover scientific events automatically.However,the key challenge is the lack of an intelligent discovery framework,involving the intelligent analysis methodology for scientific events,the intelligent verification mechanism for scientific events and the long-term storage architecture of scientific data.Based on this,we propose an intelligent management framework and its details from the view of data management to promote the intelligent scientific discovery.
作者
孟小峰
Meng Xiaofeng(School of Information,Renmin University of China,Beijing 100872)
出处
《中国科学基金》
CSSCI
CSCD
北大核心
2021年第3期419-425,共7页
Bulletin of National Natural Science Foundation of China
基金
国家自然科学基金项目(61941121,91846204)的资助。
关键词
科学数据
数据智能
数据管理
智能发现
知识融合
长期存储
scientific data
data intelligence
data management
intelligent discovery
knowledge fusion
long-term storage
