II-EN: Compute Infrastructure for Large-Scale Data Analytics

该计算研究基础设施项目旨在增强伍斯特理工学院（WPI）的高性能基础设施，以支持其在可扩展基础设施上围绕大数据分析的持续研究和教育。增强的计算集群将提供一个可扩展的基础设施，以支持沿着许多重要方向的大规模数据分析研究，包括：(1)一个成熟的软件平台，通过以流窗口为中心的查询处理，无缝地支持大规模连续和重复的工作负载；(2)可扩展的注释和来源管理，原生支持注解作为大数据的一级公民，并在查询处理和质量推理中利用其语义；(3)复杂事件分析，将复杂事件处理与OLAP技术相结合，在高速大容量数据中实现高效的多维事件模式检测；(4)基于高速数据流的多模型驱动分析系统，对衍生模型和模式进行可视化探索。这些项目共同解决了与大容量、高速度和各种数据分析相关的关键研究挑战。该基础设施预计将大大加强WPI本科和研究生水平的研究和教育的整合，特别是WPI以项目为导向的本科课程，要求每个本科生完成为期一年的研究项目。该基础设施还将使WPI不同部门的不同研究人员（就像当前的计算集群一样）受益，以进行数据密集型研究。包括研究出版物、开源软件和教育材料在内的基础设施使研究成果得以广泛传播，预计将使整个科学界受益。

This Computing Research Infrastructure project aims to enhance Worcester Polytechnic Institute's (WPI) high-performance infrastructure to support its ongoing research and education centered around big data analytics over scalable infrastructure. The enhanced compute clustering will provide a scalable infrastructure to support large-scale data analytics research along a number of important directions, including (1) a full-fledged software platform for seamlessly supporting large-scale continuous and recurring workloads with stream-window-centric query processing, (2) scalable annotation and provenance management for natively supporting annotations as first-class citizen over big data and exploiting their semantics in query processing and quality inference, (3) complex event analytics that combines complex event processing with OLAP techniques for efficient multi-dimensional event pattern detection in high-speed large-volume data, and (4) multi-model-driven analytics system on high-velocity data streams with visual exploration of the derived models and patterns. The projects collectively address key research challenges related to the analysis of large volume, high velocity, and wide variety of data.The infrastructure is expected to significantly enhance the integration of research and education at both the undergraduate and graduate levels at WPI, including in particular, WPI's project-oriented undergraduate curriculum, which requires each undergraduate to complete a year-long research project. The infrastructure will also benefit a diverse group of researchers at WPI in different departments -- as is the case with the current compute cluster -- to conduct their data-intensive research. Broad dissemination of results of research enabled by the infrastructure including research publications, open-source software, and educational materials are expected to benefit the scientific community at large.