项目名称: 大规模复杂动态图可视化关键技术研究
项目编号: No.61202279
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 计算机科学学科
项目作者: 刘真
作者单位: 杭州电子科技大学
项目金额: 23万元
中文摘要: 大规模动态网络数据的分析是实现网络的认识和干预的重要手段。面对如此大规模的复杂动态图数据,如何把获得的信息以更直观、更容易理解的方法呈现是图数据分析面临的挑战。大规模图给图可视化提出重要的研究问题,例如可视混乱、布局、导航和评价标准等。针对这些问题,本项目重点探索三方面的研究内容:首先研究大规模图数据的简化和聚类方法,降低大规模图数据规模的同时保留图的主要结构。其次,研究基于GPU和GPU集群的高扩展性并行大规模图可视布局方法。并行的多层次k-way 图划分方法获得高并行度的同时维持高划分质量。基于能量算法的图可视化布局,不但可以产生美观、可解释和可读性强的图布局,而且获得局部和全局最小能量。最后,为了克服单体显示器显示尺寸和分辨率的不足,研究大屏幕投影环境中新型适合大规模复杂动态图导航和交互的新技术。这对帮助用户贯通零散的知识发现,发掘隐含特征、潜在关联性和关系模式有重要的意义。
中文关键词: 图;关系数据;可视化;可视分析;
英文摘要: Large-scale dynamic network data analysis is an important means to achieve network awareness and intervention. Faced with such a large scale, the complexity of dynamic graph data, showing how to get information more intuitive and easier to understand the challenges graph data analysis.The scale of the graph data to the graph visualization presented important research problems, such as visual clutter, layout, navigation, and evaluation criteria. To solve these problems, the focus of this project to explore three aspects of research: first to study the simplification and clustering methods for large-scale graph data, reducing the size of the large-scale graph data, while preserving the main structure of the graph. Secondly, parallel large-scale visual layout method based on the high scalability of the GPU and GPU cluster. Parallel multi-level k-way graph partitioning method to obtain high degree of parallelism while maintaining the quality of the high divide. Graph visual layout algorithm based on energy, not only can produce beautiful, be interpreted and readability of the graph layout, and access to local and global minimum energy. Finally, in order to overcome the disadvantage of of size of resolution of single monitor, emerging new technology for large, complex dynamic map navigation and interaction in large s
英文关键词: graph;relational data;visualization;visual analysis;