非结构化数据管理（原理与实现技术）.pptVIP

Managing Unstructured Data (非结构化数据管理) 张孝(Xiao ZHANG) zhangxiao@ruc.edu.cn Agenda Motivation 9 Important Issues BUD --- Bank of Unstructured Data A Reference Framework Case Study Video Management Platform Near Future Works Motivation--Why focus on USD? BUD --- A Reference Framework Architecture layered extensible scalable distributed *DB approach DB-Arch.-like 9 Important Issues on USD fully extensible architecture universal data model the unstructured has unknown structure distributed TSM adaptive data portal evolving metadata repos. open sesame QoS understand the semantics 9 Important Issues on USD version mgmt refined vs. purging optimization imprecise information standardized unified query language standard makes easy interoperation killer application optional but Important Case #1 ---Video Mgmt Platform example scenario To recognize customers who had been the store and bought something before. Once identifying them, VMP alerts salespersons to provide them better shopping experience, eg. special offers or recommendation. Func: recognition - pattern retrieval - real-time help research topics extensible and scalable architecture hybvideo/UHS(Uninterrupted Helper System) fast knowledge(eg. behavior pattern) mining adaptive storage mgmt refinement iteratively Case #1 ---Video Mgmt Platform Basic Architecture Neare Future Works Compliance with BUD SOA Optimization feedback-based Advanced keyword search interplay of rel.+semi.+usd Many open problems to be solved just kick-off Sunday, July 24, 2010 * * 教育部数据工程与 知识工程重点实验室 School of Information, Renmin University of China Key Laboratory of Data Engineering and Knowledge Engineering, MOE Unstructured Data is value-added various potential values, NOT burden information, trend, knowledge, patterns pervasive non-formed is everywhere volume increases exponentialy 80% unstructured, incl. web data, vid/aud BUD -- a reference framework manage unstructured data extract information with time-going Thanks Discussion