1. Temporal Queries in XML Document Archives and Web Warehouses
Author: Fusheng Wang and Carlo Zaniolo
Publication: TIME-ICTL’03, IEEE
報告人：陳建宏

2. outline Main Issue Previous Work A Logical Model for Versions
Complex Queries
Conclusions

3. outline Main Issue Previous Work A Logical Model for Versions
Complex Queries
Conclusions

4. Main issue
Concisely representing the successive versions of a document as an XML document that implements a temporally grouped data model
Using XML query languages, such as XQuery, to express complex queries on the content of a particular version, and on the temporal evolution of the document elements and their contents

5. outline Main Issue Previous Work A Logical Model for Versions
Complex Queries
Conclusions

6. Temporal XML Representation
A valid tag for XML/XHTML documents is proposed to support valid time
A data model is proposed for temporal XML documents
XML query languages such as Xpath and XQuery
<valid>
<validity from=“ ” to=“ ”/>
This is test
</valid>
<title time:valid=" , ">XML</title>
<title time:valid=" , now">Introduction to XML</title>
Xquery example:
select all the title elements in the "books.xml" file
doc("books.xml")/bookstore/book/title

7. Change Detection XyDiff X-Diff Ladiff
How to know the difference between the
latest version and the previous one?
There are some of efficient change detection
algorithms
Ladiff
XyDiff
X-Diff

8. Managing the Deltas There are different schemes for storing these
deltas on secondary storage and processing
them for version reconstruction and query
execution
Revision Control System (RCS) scheme
Reference-Based Version Model(RBVM) scheme
Source Code Control System(SCCS) scheme

9. Temporal Databases
Many temporal data models are proposed already. They are classified into two main categories
Temporally ungrouped
Temporally grouped

10. outline Main Issue Previous Work A Logical Model for Versions
Complex Queries
Conclusions

11. XML REPRESENTATION of versioned document
V-Document
Each element is assigned with two attributes vstart and vend, which represent the valid version interval of the element
In V-Document, we represent each attribute by a sub-element denoted by a special flag attribute isAttr.

12. Sampled versioned xml documents
<chapter no="1">
<title>Introduction</title>
<section>Background</section>
<section>Motiviation</section>
</chapter>
</document>
version
<document>
<chapter no="1">
<title>Introduction and Overview</title>
<section>Background</section>
<section>History</section>
</chapter>
<chapter no="2">
<title>Related Work</title>
<section>XML Storage</section>
</document>
version
<document>
<chapter no="1">
<title>Introduction and Overview</title>
<section>Background</section>
<section>History</section>
</chapter>
<chapter no="2">
<title>Related Work</title>
<section>XML Indexing</section>
</document>

13. Attribute of element ”chapter”
V-document Example
Attribute of element ”chapter”
<document vstart=" " vend="now">
<chapter vstart=" " vend="now">
<no isAttr="yes" vstart=" " vend="now">1</no>
<title vstart=" " vend=" ">Introduction</title>
<title vstart=" " vend="now">Introduction and Overview</title>
<section vstart=" " vend=" ">Motivation</section>
<section vstart=" " vend="now">Background</section>
<section vstart=" " vend="now">History</section>
</chapter>
<chapter vstart=" " vend="now">
<no isAttr="yes" vstart=" " vend="now">2</no>
<title vstart=" " vend="now">Related Work</title>
<section vstart=" " vend=" ">XML Storage</section>
<section vstart=" " vend="now">XML Indexing</section>
</document>
vstart and vend tag

14. V-document advantage No storage redundancy
The changes are naturally representated
Temporal queries are easily expressed on V-documents

15. Change operations - update
<document>
<chapter no="1">
<title>Introduction</title>
<section>Background</section>
<section>Motiviation</section>
</chapter>
</document>
<title vstart=" " vend=“now">
Introduction
</title>
<title vstart=" " vend=" ">
Introduction
</title>
<title vstart=" " vend="now">
Introduction and Overview
<document>
<chapter no="1">
<title>Introduction and Overview</title>
<section>Background</section>
<section>History</section>
</chapter>

16. Change operations - insert
<section vstart=" " vend="now">
Background
</section>
<document>
<chapter no="1">
<title>Introduction</title>
<section>Background</section>
</chapter>
</document>
<section vstart=" “ vend="now">
Background
</section>
<section vstart=" " vend="now">
History
<document>
<chapter no="1">
<title>Introduction and Overview</title>
<section>Background</section>
<section>History</section>
</chapter>

17. Change operations - DELETE
<section vstart=" " vend="now">
Motivation
</section>
<document>
<chapter no="1">
<title>Introduction</title>
<section>Background</section>
<section>Motivation</section>
</chapter>
</document>
<section vstart=" " vend=" ">
Motivation
</section>
<document>
<chapter no="1">
<title>Introduction and Overview</title>
<section>Background</section>
</chapter>

18. outline Main Issue Previous Work A Logical Model for Versions
Complex Queries
Conclusions

19. Sampled versioned xml documents
<chapter no="1">
<title>Introduction</title>
<section>Background</section>
<section>Motiviation</section>
</chapter>
</document>
version
<document>
<chapter no="1">
<title>Introduction and Overview</title>
<section>Background</section>
<section>History</section>
</chapter>
<chapter no="2">
<title>Related Work</title>
<section>XML Storage</section>
</document>
version
<document>
<chapter no="1">
<title>Introduction and Overview</title>
<section>Background</section>
<section>History</section>
</chapter>
<chapter no="2">
<title>Related Work</title>
<section>XML Indexing</section>
</document>

20. Query 1 Snapshot queries: retrieve the version of the
document on :
for $e in document("V-Document.xml") /document
return snapshot( $e," “)
<document vstart=" " vend="now">
<chapter vstart=" " vend="now">
<no isAttr="yes" vstart=" " vend="now">1</no>
<title vstart=" " vend="now">Introduction and Overview</title>
<section vstart=" " vend="now">Background</section>
<section vstart=" " vend="now">History</section>
</chapter>
<chapter vstart=" " vend="now">
<no isAttr="yes" vstart=" " vend="now">2</no>
<title vstart=" " vend="now">Related Work</title>
<section vstart=" " vend="now">XML Indexing</section>
</document>

21. Query 2
Change queries: retrieve the unchanged elements between the version on and the version on :
for $e in document("V-Document.xml")
/document
return diff-identical($e," "," ")
<document vstart=" " vend="now">
<chapter vstart=" " vend="now">
<no isAttr="yes" vstart=" " vend="now">1</no>
<section vstart=" " vend="now">Background</section>
</document>

22. Query 3 Evolutionary queries: find the history of the
title of chapter 1:
for $title in document("V-Document.xml")
/document/chapter[no="1"]/title
return $title
<title vstart=" " vend=" ">Introduction</title>
<title vstart=" " vend="now">Introduction and Overview</title>

23. Query 4 Continuous queries: find titles that didn’t
change for more than 2 consecutive years.
for $title in document("V-Document.xml")
/document/chapter/title
let $duration := substract-dates(
where dayTimeDuration-greater-than(
$duration, "P730D")
return $title
<title vstart=" " vend="now">
Introduction and Overview
</title>

24. Query 5
A Since B: find chapters in which section “History” remains unchanged since the version when the title was changed to “Introduction and Overview”.
for $ch in document("V-Document.xml")
/document/chapter
let $title := $ch/title[
title="Introduction and Overview"]
let $sec := $ch/section[section="History"]
where not empty($title) and not empty($sec)
and =
and ="now"
return $ch
<chapter vstart=" " vend="now">
<no isAttr="yes" vstart=" " vend="now">1</no></chapter>

25. Query 6
A Until B: find chapters in which the title didn’t change until a new section “History” was added
for $ch in document("V-Document.xml")
/document/chapter
let $title := $ch/title[1]
let $sec := $ch/section[section="History"]
where not empty($title) and not empty($sec)
and =
return $ch
<chapter vstart=" " vend=“ ">
<no isAttr="yes" vstart=" " vend="now">1</no>
</chapter>

26. Join Query Joins: find the chair of employee ’Bob’ on 1996-01-31:
for $e in document("employees.xml")
/employees/employee[name=’Bob’]
for $d in document("depts.xml")/depts/dept
let $et := snapshot($e," ")
let $dt := snapshot($d," ")
where not empty($et) and not empty($dt)
and $et/dept = $dt/name
return $dt/chair

27. outline Main Issue Previous Work A Logical Model for Versions
Complex Queries
Conclusions

28. Conclusions We can represent a multi-version document in V-Document
In this representation scheme, complex queries can be expressed easily