A Cluster Infrastructure to Support Retrieval, Management and Visualization of Massive Amounts of Data

本项目为数据密集型应用开发基础设施，旨在建立一个集群和显示墙基础设施，以促进在广泛场景下支持海量数据的检索、查询、建模和可视化的系统实验研究。该基础设施由具有快速网络基础设施的图形工作站集群、高分辨率显示墙和专用文件服务器组成，支持三个研究项目：开发隐藏网络、大型科学可视化和大型3D环境建模和渲染。第一个项目搜索、索引和查询隐藏的Web数据。通常，通过超链接访问页面的爬虫如果不搜索各个站点，就无法找到所需的信息。该项目开发了一些技术，使爬虫能够在隐藏的页面中导航，构建了一个搜索引擎，允许查询这些数据。第二个与可视化技术的发展有关，这种技术适用于非常大的科学数据集。这项研究包括科学可视化的核心外算法的发展，并行图形和可缩放显示技术，新的渲染算法，以及更好的可视化建模原语。该项目期望构建可视化系统，有效地将可视化需求与可用硬件资源解耦，允许在任何合理的设备上对任意大数据集进行可视化。第三个项目着眼于现实世界对象和场景的模型，娱乐、训练和模拟、特效、法医记录分析、远程呈现和远程演练等应用程序都可以从中受益。这项工作需要解决与注册和清理原始数据相关的昂贵优化问题，以及用于实时渲染的高分辨率显示。这个基础设施使可视化和几何计算和WebDB实验室的建立成为可能；这反过来又会使许多学生受益。外联部分已经到位；该研究目前支持一些女性和少数族裔。此外，目前正在开设两门新的有关课程

This project, developing an infrastructure for data intensive applications, aims at establishing a cluster and display-wall infrastructure to facilitate experimental research in systems that support retrieval, querying, modeling, and visualization of massive amounts of data in a broad range of scenarios. The infrastructure, consisting of a cluster of graphics-capable workstation with fast network infrastructure, a high-resolution display wall, and a dedicated file server, supports three research projects: Exploiting the Hidden Web, Large Scientific Visualization, and Modeling and Rendering Large 3D Environments.The first project searches, indexes, and queries the hidden Web data. Often crawlers, which visit pages by following hyperlinks, cannot find needed information without searching the individual sites. This project develops techniques to enable a crawler to navigate through hidden pages, building a search engine that allows these data to be queried. The second relates to the development of visualization technology that is suitable for very large scientific datasets. This research includes the development of out-of-core algorithms for scientific visualization, parallel graphics and scalable-display technology, novel rendering algorithms, and better visualization modeling primitives. The project expects to build visualization systems that effectively decouple the visualization requirements from the available hardware resources, allowing visualization of arbitrarily large datasets on any reasonable device. The third project looks into models of real-world objects and scenes from which applications such as entertaining, training and simulation, special effects, analysis of forensic records, telepresence, and remote walkthroughs can benefit. This work requires tackling expensive optimization problems related to registering and cleaning raw data, and high-resolution display for real-time rendering.This infrastructure enables the establishing of a lab in Visualization and Geometric Computing and WebDB; which in turn will benefit many students. An outreach component is in place; the research currently supports several female and minorities. Moreover, two new related courses are under development