Conceptual Modeling for XML Data

1. Conceptual Modeling for XML Data Tok Wang Ling National University of Singapore DASFAA’2003 Panel Discussion

2. Outlines • Why do we need conceptual modeling? • What are the important semantic information to be captureｄ? • Uses of conceptual model for some XML research topics

3. Motivation: Why do we need to have a conceptual model to represent XML Data? <department number =“cs”> <name> computer science</name> <course number = “cs4221”> <name> Database </name> <student number = “1234” > <name> B.Y.Smith</name> <grade> 70</grade> </student> <student number=“1235”> <name> C.U.Brown </name> <grade> 60</grade> </student> </course> </department> <! ELEMENT department (name,course+)> <! ATTLIST department number ID #REQUIRED> <! ELEMENT course (name, student*)> <! ATTLIST course number ID #REQUIRED> <! ELEMENT student (name, grade?)> <! ATTLIST student number CDATA #REQUIRED> <! ELEMENT name (#PCDATA)> <! ELEMENT grade (#PCDATA)> (b) An XML DTD for (a) (a) XML document

4. department &1 number name course course cs computer science &3 &2 name student number number name student student &7 &21 &22 &20 &4 &5 &6 ♦number Database ♦name ▼♦department cs4221 ▼ ♦course number ♦number name grade ♦name grade name number ▼♦student ♦number ♦name ♦grade &27 &26 &28 &25 &23 &24 C.U. Brown 1235 60 1234 B.Y.Smith 70 Motivation (cont.) (B) Dataguide (a) OEM Diagram Figure 1: Sample instance demonstrating OEM and dataguide

5. Motivation (continue) Q: What are the important semantic information and constraints cannot be captured by the DTD and Dataguide? • What are the object classes? department, course, student? • Attributes of object classes? • Identifiers of object classes? • What are the relationship types defined among object classes? e.g. Relationship types among department, course, student? • What is “grade”? Object class? Attribute of student? • Are there redundancies?

6. Semantic Information to be captured by an XML conceptual model • Object class • attributes of object class • ordering on object class • Relationship Type • Represent hierarchical structure • degree of n-ary relationship type • participation constraints of object classes in relationship type • attributes of relationship type • disjunctive relationship type • recursive relationship type • Reference

7. Semantic Information to be captured by an XML conceptual model (cont.) • Attribute • key attribute / identifier • composite attribute • disjunctive attribute • attributes with unknown structure • fixed and default values of attribute • derived attribute • Functional dependencies and other constraints • Inheritance hierarchy (class hierarchy) • Semi-structured data instance representation

8. Q: What are the semantic information cannot be represented by Dataguide, DTD, XML Schema? • Attribute or object class • Degree of relationship type • Attibute of object class or relationship type • Class hierarchy • Functional dependency • …

9. department 2, 1:n, 1:1 number name course cs student number name cs number name grade A solution: ORA-SS, an object-relationship attribute model for semi-structured data. department Number: CS Name: Computer Science course course student student Number student Name Number: CS4221 Name: Database number: 1235 Name: C.U. Brown Grade: 60 number: 1234 Name: B.Y. Smith Grade: 70 Figure 2: ORA-SS instance diagram Figure 3: ORA-SS schema diagram

10. project member publication p1 pub1 m1 pub2 p2 m2 p3 pub3 project ▼♦project 2, +, + ♦ id ♦ name ▼♦ member member ♦ name id name ♦ job title 2, *, + ▼♦ publication ♦ number name job title ♦title publication number title The data model of ORA-SS - Relationship Type • attributes of relationship type • degree of n-ary relationship type • participation constraints of objects in relationship type • disjunctive relationship type • recursive relationship type (c) Dataguide (a) ORA-SS Schema Diagram (b) Instance Relationship Figure 5: Representing binary relationship type

11. project member publication p1 pub1 m1 pub2 p2 m2 p3 pub3 project ▼♦project 2, +, + ♦ id ♦ name ▼♦ member member ♦ name id name ♦ job title 3, *, + ▼♦ publication ♦ number name job title ♦title publication number title The data model of ORA-SS - Relationship Type (cont.) (b) Instance Relationship (a) ORA-SS Schema Diagram (c) Dataguide Figure 6: Representing ternary relationship type

12. course cs 2, 4:n, 3:8 code * student title ANY cs cs dept prefix D: comp number * number first name last name mark grade The data model of ORA-SS - Attribute • key attribute • composite attribute • disjunctive attribute • attribute with unknown structure • fixed and default values of attribute • derived attribute hobby Figure 7: Object classes with relationship type and attributes in an ORA-SS schema diagram

13. Uses of the Conceptual model for XML research • Normal form XML schema • remove redundant data • resolve multiple inheritance conflicts • Storage structure for XML databases • use Object Relational Model • XML Views • derived information from references and class hierarchy • defining views • materialized view maintenance • view updates • Integration of XML documents • Evaluating XML queries on XML databases

14. professor 2 staff# name course 2 C.Code title textbook + title author ISBN Research Topics using ORA-SS ModelNormal Form XML Schema • Schema may have a lot of redundant data • Update anomalies • Normal Form schema is needed

15. professor course textbook C.Code name title staff# + courseR textbookR ISBN author title professor course textbook title name C.Code + ISBN title author professorR textbookR Research Topics using ORA-SS ModelNF XML Schema (cont.) • Some better solutions: • Redundancies are removed, in normal form textbook-Ref course-Ref textbook-Ref professor-Ref staff#

16. Research Topics using ORA-SS ModelStorage Structure for XML Databases • Main Rules • Each object class together with its attributes form a nested relation (object relation) • Each relationship type together with its attributes form a nested relation (relationship relation) • Nested relations can be handled by Object Relational model, e.g. ORACLE 8i.

17. Supplier SP 2 + Name City S# SP Price Part SPJ 3 P# Name Color Project SPJ J# Name Loc Qty Research Topics using ORA-SS ModelStorage Structure for XML Databases Object Relations Supplier (S#, Name, (City)) Part (P#, Name, color) Project(J#, Name, Loc) Relationship relations SP (S#, P#, price) SPJ (S#, P#, J#, Qty) Constraint: SPJ[S#, P#]  SP[S#, P#]

18. course faculty course cs, 2,4:n,3:8 cs fs, 2,1:n,1:n faculty name title code title code student student studentR cs cs fs | grade faculty grade history | student number * grade student number hostel home hostel home grade course code street number street name number name Research Topics using ORA-SS ModelXML Views • What information can be directly derived from references and class hierarchy Referencing an object class in an ORA-SS schema diagram

19. Research Topics using ORA-SS ModelXML Views (cont.) • Valid views of an ORA-SS schema • Operations: selection, projection, join, up/down Project Project Part Supplier 2 2 2 Part Part Supplier Part 2 2 2 3 3 2 3 price price Supplier Project Project total_qty 2 3 3 3 price Qty Qty Qty View 1 View 2 View 3

20. Conclusion • A good conceptual model is needed for XML database applications: * normal form schema * storage structure * view design and view updates * ….