Semantic Data Modeling using XML Schemas

1
Semantic Data Modeling using XML Schemas

• Murali Mani, Dongwon Lee,
• Prof. Richard R Muntz
• UCLA
• Nov 27, 2001

2
Outline of the presentation

• Data modeling using XML Schemas why do we need
  to study this and how difficult is it?
• XGrammar characteristics and notations
• XGrammar ? Entity relationship model
• Some interesting issues

3
Motivation

• Data modeling is very crucial in successful
  database design.
• Significantly new features provided by XML
  Schemas
• Ordered relationships

4
Ordered relationships
(0, )
(1, )
Book
Person
author
Book (0, ) gt Person (1, )
5
Motivation (contd)

• Data modeling is very crucial in successful
  database design.
• Significantly new features provided by XML
  Schemas
• Ordered relationships
• Closure under union

6
XGrammar

• Defined by a 7-tuple (NT, NH, T, S, E, H, A)
• NT is the set of tree types
• NH is the set of hedge types
• T is the set of terminal symbols
• S is the set of start symbols, S ? NT
• E is the set of element production rules
• H is the set of hedge production rules
• A is the set of attribute production rules

7

• NT Library, Book, Magazine, Person
• NH LibraryContent
• S Library
• E Library ? library (LibraryContent)
• Book ? book (?)
• Magazine ? magazine (?)
• H LibraryContent ? (Book, Magazine)
• A Book ? book (_at_authorsIDREFS Person)
• Magazine ? magazine (_at_editorIDREF
  Person)

8
XGrammars are closed under union

• Union of any two tree types is a tree type
• Book ? book (Title, Author)
• Magazine ? magazine (Name, Editor)
• ReadingMaterial ? (Book Magazine)
• Schema language proposals such as RELAX, TREX,
  RELAX-NG, XDuce are closed under union.

9
Mapping from EER ? XGrammar

• Using Element Production rules ordered 1n
  binary
• relationships

(1, 1)
(1, )
Book
Person
author
Book (1, 1) gt Person (1, )
E Book ? book (Person)
10
2. Using IDREF unordered 1n binary
relationships
(1, 1)
(0, )
Book
Person
author
Book (1, 1) ? Person (0, )
A Person ? person (_at_authorOfIDREF
Book)
11
3. Using IDREFS ordered nm binary
relationships
(0, )
(1, )
Book
Person
author
Book (0, ) gt Person (1, )
A Book ? book (_at_authorIDREFS Person)
12
Significant challenges
Library
(1, 1)
(1, 1)
has
has
(0, )
(0, )
Book
Article
(0, 1)
(0, 1)
Person
author
editor
(1, )
(1, )
13
Best XGrammar possible at present

• E Library ? library (Book, Article, Person)
• A Book ? book (_at_authorsIDREFS
  Person)
• Article ? article (_at_authorsIDREFS
  Person)

14
Desired XGrammar

• E Library ? library (Book, Article)
• Book ? book (Person)
• Article ? article (Person)
• We need more information for this
• PersonBook ? PersonArticle ?
• PersonBook ? PersonArticle Person