Doan Dai Duong and Le Thi Thu Thuy

1
A Unified Framework for the Semantic Integration
of XML Databases
First  IEEE International Conference on Digital
Information Management (ICDIM)

• Doan Dai Duong and Le Thi Thu Thuy
• Duong_Dai.Doan, Thuy_Thi_Thu.Le_at_unb.ca The
  University of New Brunswick, Fredericton, NB,
  Canada

Presented by Virendrakumar C. Bhavsar
December 06-08, 2006
2
Agenda

• Introduction
• XML Declarative Description (XDD)
• Modeling of Data Components
• Modelling of Processing Components
• Conclusion

3
Introduction

• General model of XML database integration

Step 1 Schema Integration
4
Step 2 Query Processing
Users
4
5
Proposed Integration Framework

• Powerful
• XDD supports for all tasks of framework
• Input XML query, input XML data, output XML data
• Rules, constraints, mappings
• Metadata
• Based on XML standard format, XDD combines all
  tasks of framework tightly and makes it easily to
  manipulate data
• Reduce time and effort of programmers and users
  and syntax errors

5
6
XML Declarative Description

• XML Declarative Description (XDD) is XML-based
  information representation
• Ordinary XML expressions (ground XML
  expressions) variables Non-ground XML
  expressions
• ? Enhancement of expressive power and
  representation of implicit information
• XML clauses of the form
• H ? B1, , Bm, C1, , Cn
• ? Able to express conditions, constraints

Wuwongse, V., Anutariya, C., Akama, K., and
Nantajeewarawat, E. XML Declarative Description
(XDD) A Language for the Semantic Web. IEEE
Intelligent Systems, Vol. 16, No. 3, (2001) 54-65
7
Modeling of Data Components

• XML Databases
• Extension (actual data values) ground XML
  expressions
• Intension (schemas, logical specifications,
  relationships, indexes and constraints)
  non-ground XML expressions
• XML Queries
• Include constructor, patterns, and filters
• Correspond to three parts (H, Bi, Cj) of XDD rule
  
• H ? B1 , Bm, C1,,Cn

8
Modeling of Data Components
Query modelled by XDD
9
Query Execution Example
10
Modeling of Data Components

• Mappings
• Describes correspondence between object in
  integrated schema and its corresponding objects
  in local schemas
• Supports decomposing XML queries and converting
  data
• Modeled by non-ground XML expressions

11
Sample of Mappings
Object in integrated schema
Object in schema A
Object in schema B
12
Modelling of Processing Components

• Schema Integration Component
• The main task is to resolve conflicts between
  schemas of participating databases
• Conflict resolution between various schemas is
  done at one time (one-shot strategy)
• Each local schema is big non-ground XML
  expression (E_variable)

13
Schema Integration Component

• XDD can interactively process all schemas as E
  expressions

14
Schema Conflict Classification
Conflicts between schemas can be classified into
four main kinds

• Naming conflicts
• Synonyms
• Acronyms
• Homonyms
• Structural conflicts
• Missing items conflicts
• Internal path discrepancy conflicts
• Aggregation conflicts
• Generalization/specification

• Constraint conflicts
• Occurring numbers of elements
• Fixed vs. default values
• Constraints of attributes
• Data type conflicts
• Disjoint or incompatible data types
• Compatible data types
• IDREF and IDREFS

15

• Aggregation
• conflict

14
16
Query Decomposition

• The main task ? yield n local subqueries from
  global query

ltstudent id Sidgt ltnamegtSnamelt/namegt ltcountr
ygtScountrylt/countrygt lt/studentgt
Integrated schema
ltSATstudent key Sid sourceBgt ltfullnamegt
Sname lt/fullnamegt ltcountrygtScountrylt/countrygt
lt/SATstudentgt
Schema for source B
ltSOMstudent idSid sourceAgt ltnamegt Sname
lt/namegt ltnationgtScountrylt/nationgt lt/SOMstudentgt
Schema for source A
17
Query Decomposition
Mappings from global to local
A. Brief view
Sub query for local source
query
Query Decomposition
Sub query for local source

• B. Solution

Input XML query
ltstudent id Sidgt ltnamegtSnamelt/namegt ltcountr
ygtScountrylt/countrygt lt/studentgt
16
18
Query Decomposition Example
ltanswergt ltSATstudent sourceBgt
ltcountrygtScountrylt/countrygt lt/SATstudentgt ltSOMst
udent sourceAgt ltnationgtScountrylt/nationgt
lt/SOMstudentgt lt/answergt
Local query for source A
results in
4
Local query for source B
ltanswergt Eexpression lt/answergt ?
ltMappinggt ltstudentgt
ltcountrygtScountrylt/countrygt lt/studentgt
ltlocalgtEexpressionlt/localgt lt/Mappinggt
3
infers to
matches with
1
ltMappinggt ltstudentgt ltcountrygtScountrylt/coun
trygt lt/studentgt ltlocalgt ltSATstudent
sourceB"gt ltcountrygtScountrylt/countrygt lt/
SATstudentgt ltSOMstudent sourceA"gt
ltnationgtScountrylt/nationgt lt/SOMstudentgt
lt/localgt lt/Mappinggt
bounds to
2
19
Query Decomposition

• Using special structure of mapping and applying
  XDD rules for query decomposition
• Subqueries for distributed data sources are
  simultaneously produced
• Similarly for data conversion, extracted data are
  simultaneously converted to global schema format
  

20
Conclusion

• XDD is used to model all data components and
  processing components of XML database integration
  framework
• Components of system modeled by XDD can
  communicate and exchange data easily
• Special structure for XDD-based bidirectional
  mappings is designed. Information is produced
  efficiently for both query decomposition and data
  conversion, avoiding data redundancy
• The framework can
• Integrate n participating schemas
• Decompose a query into n subqueries at a time.

21
Thank you !