大数据可视分析综述

摘要

Visual analytics is an important method used in big data analysis. The aim of big data visual analytics is to take advantage of human’s cognitive abilities in visualizing information while utilizing computer’s capability in automatic analysis. By combining the advantages of both human and computers, along with interactive analysis methods and interaction techniques, big data visual analytics can help people to understand the information, knowledge and wisdom behind big data directly and effectively. This article emphasizes on the cognition, visualization and human computer interaction. It first analyzes the basic theories, including cognition theory, information theory, interaction theory and user interface theory. Based on the analysis, the paper discusses the information visualization techniques used in mainstream applications of big data, such as text visualization techniques, network visualization techniques, spatio-temporal visualization techniques and multi-dimensional visualization techniques. In addition, it reviews the interaction techniques supporting visual analytics, including interface metaphors and interaction components, multi-scale/multi-focus/multi-facet interaction techniques, and natural interaction techniques faced on Post-WIMP. Finally, it discusses the bottleneck problems and technical challenges of big data visual analytics.%可视分析是大数据分析的重要方法。大数据可视分析旨在利用计算机自动化分析能力的同时，充分挖掘人对于可视化信息的认知能力优势，将人、机的各自强项进行有机融合，借助人机交互式分析方法和交互技术，辅助人们更为直观和高效地洞悉大数据背后的信息、知识与智慧。主要从可视分析领域所强调的认知、可视化、人机交互的综合视角出发，分析了支持大数据可视分析的基础理论，包括支持分析过程的认知理论、信息可视化理论、人机交互与用户界面理论。在此基础上，讨论了面向大数据主流应用的信息可视化技术--面向文本、网络(图)、时空、多维的可视化技术。同时探讨了支持可视分析的人机交互技术，包括支持可视分析过程的界面隐喻与交互组件、多尺度/多焦点/多侧面交互技术、面向 Post-WIMP 的自然交互技术。最后，指出了大数据可视分析领域面临的瓶颈问题与技术挑战。