BEGIN:VCALENDAR
PRODID:-//hacksw/handcal//NONSGML v1.0//EN
METHOD:PUBLISH
BEGIN:VEVENT
DESCRIPTION:Click for Latest Location Information: http://semtech2011.semanticweb.com/sessionPop.cfm?confid=62&proposalid=3851\nFor large-scale semantic database applications that involve complex queries and/or complex inferencing, the huge memories and parallel computing horsepower of supercomputers provide a significant advantage. The authors of this presentation have been exploring approaches to implementing semantic web databases on the Cray XMT supercomputer, a parallel system that can be configured with several terabytes of globally addressed, shared memory. We will describe some of our research results to date, including:\n<ul><li>Analyzing a full day?s worth of Twitter data, rapidly identifying the tree-structured news-dissemination portion as well as the conversation clusters.\n<li>Performing RDFS closure on a billion-triple data set, around 7x faster than previous results.\n<li>Performing prototype queries on a data set of 20 billion triples (all held in main memory), while scaling up to 512 processors.\n<li>Prototyping an API designed to facilitate porting commercial semantic web databases onto the XMT.</ul>
DTSTART:20110607T134000
SUMMARY:Semantic Databases and Supercomputers
DTEND:20110607T142959
LOCATION: See Description
END:VEVENT
END:VCALENDAR