基于属性偏序结构图的文本型灾情多元信息可视化

摘要

在处理和分析文本型灾情多元信息时，由于传统概念格层次结构不够清晰，不利于对其进行数据挖掘和知识发现以支持决策。应用形式概念分析理论，在分层概念格建格算法的基础上，通过形式背景行列交换原理，对形式背景进行优化，提出了属性偏序结构图表示方法。该方法可以实现层次化的属性聚类，便于分析概念构成，达到分层递阶可视化的效果，实现知识发现和多元海量数据的信息融合。应用该方法，绘制了中国1995-1999年农作物受灾和成灾情况的属性偏序结构图。结果表明，该方法清晰、直观，可用于支持文本型灾情信息的分析和趋势判断。%An hierarchy of traditional concept lattice is not clear in processing and analyzing multivariate in-formation about textual disaster.This is not conducive to carry on data mining and knowledge discovery to support decision making.It is proposed that the representation method of attribute partial order structure diagram by apply-ing the ranks exchange principle of formal context to optimize formal context.The method has advantage in analy-zing some concepts and hierarchical attribute clustering.It can realize knowledge discovery,hierarchical informa-tion visualization and information fusion of huge amounts of multivariate data.The method was applied to draw mul-tiple attribute partial order structure diagrams of the national crop disaster and inundated area during 1995 -1999. The results show that this approach is clear,intuitive,and supportiveto information analysis and trend judgment of textual disaster information.