图表大数据解析方法综述  

Review of Parsing Methods for Big Data in Chart

作　　者：容姿 丁一 李依泽 周志光 Rong Zi;Ding Yi;Li Yize;Zhou Zhiguang(Big Data Visualization and Human Computer Collaborative Intelligent Laboratory,Hangzhou Dianzi University,Hangzhou 310018)

机构地区：[1]杭州电子科技大学大数据可视化与人机协同智能实验室,杭州310018

出　　处：《计算机辅助设计与图形学学报》2025年第2期216-228,共13页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(62277013);之江实验室开放课题(K2022KG0AB01);全国统计科学研究项目(2022LY099);浙江省教育科学规划重点课题(GZ2023112);浙江省普通本科高校“十四五”教学改革项目(jg20220214);浙江大学计算机辅助设计与图形系统全国重点实验室开放课题(A2301)。

摘　　要：图表以视觉逻辑的形式有效地传达信息,为理解关系隐晦的复杂数据提供了直观的途径,在科学文档中扮演着至关重要的角色.然而,图表本身的视觉结构丰富性和视觉语义复杂性为图表可视化的构建提出了挑战,而底层数据的内在多样性和用户认知视角的差异性又给图表信息的解读带来了分歧.随着自动图表解析技术的兴起,提供了一种准确、有效的途径来解读图表,为更高级的图表洞察和推理奠定了基础,并进一步辅助决策制定.文中对图表大数据解析方法的相关研究进行综述,首先从研究对象图表本身和用户两端出发,阐明图表解析过程的必要性和复杂性;然后按照图表感知、结构解析和洞察推理3个层次,分别概述图表解析过程的研究方法和前沿技术;进一步,阐述图表解析的下游应用,如检索、交互、生成等;最后总结图表解析方法在理解复杂图表类型和保证信息完整性等方面面临的挑战,并对其未来的发展趋势进行展望.Charts effectively convey information through visual logic,providing an intuitive pathway for understanding complex data with obscure relationships and playing a crucial role in scientific documenta-tion.However,the richness of a chart’s visual structure and the complexity of its visual semantics present challenges in creating chart visualizations,while the inherent diversity of underlying data and differences in users’cognitive perspectives bring divergences in interpreting chart information.With the emergence of automatic chart parsing technologies,an accurate and effective means has been established to interpret charts,laying the groundwork for more advanced insights and reasoning,and further facilitating decision-making.This paper provides a review of research on parsing methods for big data of Chart.It starts by elucidating the necessity and complexity of the chart parsing process from both the perspectives of the chart itself and users.It then outlines the research methods and cutting-edge technologies in chart parsing at three levels:chart perception,structural parsing,and insight reasoning.Furthermore,it elaborates on the down-stream applications of chart parsing,including retrieval,interaction,and generation.Finally,the paper summarizes the challenges faced by chart parsing methods,particularly in understanding complex chart types and ensuring information completeness,and presents prospects for future developments in the field.

关 键 词：图表解析 图表提取 图表信息解读 可视化分析 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]