MIX A Meta-Data Indexing System for XML

Document Sample
MIX A Meta-Data Indexing System for XML Powered By Docstoc
					MIX: A Meta-Data Indexing
System for XML

   SungRan Cho, L3S
   Nick Koudas, University of Toronto
   Divesh Srivastava, AT&T Labs-Research
Motivation

   Diverse information over the Web
    - quality (e.g., accuracy, recency), security

   Queryable for meta-data with underlying data

   Annotate XML data with a variety of meta-data
   Efficiently identify data that meet desired constraints on
    meta-data
Our Meta-data Model
   Meta-data levels
    - Assigned to elements by associating an attribute

   One-sided range constraint
    - find all sales with approximation error < 5%
    - update timestamp > 2005-Jan-01
    - find senate documents with security level < `secret`

   Inheritance
    - A meta-data level is automatically inherited by its sub-
     elements unless explicitly specified
MDI (Meta-data Indexes)
   Develop a family of Meta-data Indexes
    - Multi-dimensional index structure

   Features of Meta-data Indexes
    - Actual meta-data levels
    - Inherited meta-data levels
    - Inheritance source nodes
MIX (Meta-data Indexing for XML)

   Meta-data indexed querying
    - Support for all XPath axes

   Dynamic maintenance of meta-data
    - Add, delete, change
Architecture of MIX

   MDI Management                       XPath
                                        Querying
     Meta-data
     Specification
                                         Meta-data
                                         threshold
     Bulk Load             XML Stores
                                          Execution
  Index       Meta-data   MDI             Engine
  Maintenance change

                                        Query Result
Thank you