农业大数据学报 ›› 2022, Vol. 4 ›› Issue (2): 108-118.doi: 10.19788/j.issn.2096-6369.220216

• 专题——科研大数据 • 上一篇    

防返贫监测大数据融合框架的构建与应用

王鑫(), 郭雷风()   

  1. 中国农业科学院农业信息研究所,北京 100081
  • 收稿日期:2022-05-06 出版日期:2022-06-26 发布日期:2022-11-08
  • 通讯作者: 郭雷风 E-mail:30103398@qq.com;guoleifeng@caas.cn
  • 作者简介:王鑫,男,博士研究生,研究方向:农业农村信息化; E-mail:30103398@qq.com
  • 基金资助:
    中央公益性科研院所基本科研业务费专项“数字经济视角下脱贫地区巩固拓展脱贫攻坚成果同乡村振兴有效衔接的机制与路径研究”(JBYW-AII-2021-38)

Application and Construction of Big Data Fusion Framework for Anti-poverty Monitoring: A Systematic View of Data, Models, and Applications

Xin Wang(), Leifeng Guo()   

  1. Agricultural Information Institute of Chinese Academy of Agricultural Sciences, Beijing 100081
  • Received:2022-05-06 Online:2022-06-26 Published:2022-11-08
  • Contact: Leifeng Guo E-mail:30103398@qq.com;guoleifeng@caas.cn

摘要:

数据驱动的科学决策能力在乡村全面振兴背景下防返贫治理中起着越来越重要的作用。通过加强防返贫监测大数据融合问题的研究,以数据融合为核心的防返贫治理科学决策体系能够得到有效的建立。防返贫监测大数据融合关键问题有三个。第一是原始数据分析。防返贫监测大数据的原始数据非常复杂。由于返贫测量标准的多维化、返贫致贫因素的多元化等四个方面原因,导致了其数据来源具有多行业部门和多专业领域,其数据特征具有多尺度、多源异构的时空大数据的复杂特点;第二是数据融合框架建立。从理论上来说,防返贫监测大数据的数据融合框架包括数据来源、监测目标、数据种类、融合模型、数据融合应用五个层次,从而以数据驱动科学决策的视角构建了数据融合的整体框架;第三是数据融合应用。在防返贫监测与帮扶全过程中,通过数据融合,防返贫监测大数据能够为防返贫监测家庭用户画像与知识图谱、瞄准对象的识别与预测、精准帮扶策略的设计、脱贫时间预测与动态退出评估等四大方面和九项具体目标需求提供科学决策辅助。上述研究成果系统地涵盖了防返贫监测大数据融合框架中的数据、模型和应用,并在总结现有研究成果基础上进一步创新性地系统提出了防返贫监测大数据融合框架。

关键词: 防返贫, 大数据, 数据融合, 数据驱动, 科学决策

Abstract:

Data-driven scientific decision-making ability is playing an increasingly important role in the prevention of returning to poverty in the context of rural revitalization. By stepping up research on data fusion issues in this area, in the big data of anti-poverty monitoring, the scientific decision-making system for preventing poverty return linked by data fusion model can be effectively established. There are three key problems in big data fusion of anti-poverty monitoring. The first is raw data analysis. The raw data of big data for anti-poverty monitoring is very complex. Due to the multi-dimension of the measurement standard of returning to poverty, the diversification of returning to poverty factors and other four aspects. As a result, its data sources have multiple industry sectors and multiple fields of expertise, and Its data features are spatio-temporal big data with multi-scale and multi-source heterogeneity. The second is the establishment of data fusion framework. In theory, the data fusion framework of big data for anti-poverty monitoring includes five levels: data source, monitoring target, data type, fusion model and data fusion application, thus, the framework of data fusion is constructed from the perspective of data-driven scientific decision. The third is data fusion application. In the whole process of monitoring and helping poverty prevention, through data fusion, big data of anti-poverty monitoring can provide scientific decision-making assistance for four aspects and nine specific target needs, including family user portraits and knowledge maps, identification and prediction of targeted objects, design of precise help strategies, prediction of poverty alleviation time and dynamic exit evaluation. The above research results systematically cover the data, models and applications in the big data fusion framework for anti-poverty monitoring, and further creatively and systematically put forward the big data fusion framework for anti-poverty monitoring on the basis of summarizing the existing research results.

Key words: anti-poverty, big data, data fusion, data driven, scientific decision

中图分类号: 

  • G203