1. Introduction to Web Science
Maintaining Information on the Web

2. Six challenges of the Knowledge Life Cycle
Acquire
Model
Reuse
Retrieve
Publish
Maintain

3. Prediction
Ontology maintenance will become the significant problem as ontologies become more mainstream
High quality ontologies are easier to maintain

4. What makes a good quality ontology?
Coverage
Correctness
Consistency
Richness
Organisation
Relation to reality
Making meaning clear

5. Common modelling pitfalls
Instantiation
Constitution
Composition
Disjunction
Polysemy

6. Instantiation (1)
Which one is correct?
PDA
IPAQ
PDA
My PDA

7. Instantiation (2) Question Answer What PDA model do you sell?
My PDA is not a PDA model!
So you wouldn’t sell your PDA!

8. Composition (1) What’s incorrect with this? Computer Disk Drive Memory
Micro Drive
Memory

9. Composition (2) Question Answer What computers do you sell?
Definitely not disk drives and memory is the answer!

10. Disjunction (1)
Computer Part
Disk Drive
Micro Drive
Memory

11. Disjunction (2) Computer Has part Computer Part Camera Has part
Flashcard 110

12. Disjunction (2) Computer Has part Computer Part Camera Has part
Flashcard 110

13. Disjunction (3) Computer Has part Computer Part Camera Has part

14. Polysemy (1) Words with more than one meaning … What is a book?
A physical object
“The DaVinci Code”
An abstract entity
A Book by Dan Brown

15. Polysemy (2) Book The DaVinci Code Dan Brown Book
We must be consistent with what is a book!
Book
The DaVinci Code
Dan Brown Book

16. Constitution (1)
Computer should not form part of the hierarchy, different level of constitution
Entity
Matter
Metal
Clay
Physical Object
Computer

17. Constitution (2) What kind of matter conducts electricity?
Answer should not be computer

18. Making Meaning Clear Part-of relates parts to their wholes
E.g. part-of(engine,car)
Part-of is irreflexive
Part-of is anti-symmetric
Nothing can have only one part

19. Requirement for Ontology Management(1/2)
Stored, aligned, evolution need to be managed
Alignment
the ontologies either need to be integrated
Because, re-used
Ontology alignment is very relevant in a Semantic Web context
separate ontologies should be aligned and linked

20. Requirement for Ontology Management(2/2)
Support for evolving ontologies
change in the real world (E.g. Car engines)
adaptations to different tasks (E.g. Wine and food)
alignments to other ontologies (E.g. Travel and food)
Ontology library systems
grouping, reorganizing ontologies for further re-use
integration, maintenance, mapping, versioning

21. Aligning Ontologies (1)
Why is Aligning Needed
created to link different terminologies (E.g. books)
modeling styles used in these domain specific ontologies
creating bridges between separated pieces of knowledge

22. Aligning Ontologies (2)
Aligning Annotated XML Documents
(The alignments)

23. Supporting Ontology Change (1)
Ontologies specific
Not static, evolution is need
Reasons of changes in ontologies
Changes in the domain
Change in the real world
Domain to be modified
Changes in the conceptualization
Different perspective
E.g. Pet shop vrs Reptile Ontology
Changes in the specification
Translation – retain the semantics

24. Supporting Ontology Change (2)
Changes in ontologies involve Several Problems
Effects of Ontology Change
Different interpretation or may use unknown terms
Changes to the source ontology
Applications that use the ontology may also be hampered

25. Supporting Ontology Change (3)
Change Management
Comparing Ontologies
Logical definition change
Non-logical change
Identifier change
Addition of definitions
Deletion of definitions
Specification of Change
Transformation or actual change
Conceptual relation
Descriptive meta-data like date, author, and reason of the update
Valid context

26. Organising Ontologies (1)
Different ontologies increasing
Task of storing
Maintaining and re-organizing
Ontology library system
Tool in grouping and re-organizing ontologies
Re-use, integration, maintenance and versioning

27. Organising Ontologies (2)
Sesame
Open storage, identification and versioning of ontologies
Smooth access to existing ontologies and advanced support in adapting ontologies to certain domain and task-specific circumstances
Fully employing the power of standardization and providing access to upper-layer ontologies and standard representation languages

28. Organising Ontologies (3)
Sesame Requirements
(Aspects of an ontology library system)

29. Organising Ontologies (4)
Functionality of an Ontology Storage System
Management
Storage
easily accessible
Ontologies classified
Stored in modules (because ontology re-use)
Identification
Unique identifier
Versioning
Ensuring consistency

30. Organising Ontologies (5)
Functionality of an Ontology Storage System
Adaptation
Searching
how to search ontology
Editing
how to add, delete and edit specific ontologies
Reasoning
How to derive consequences from an ontology

31. Organising Ontologies (6)
Functionality of an Ontology Storage System
Standardization
Language
Kind of standard ontology language used in the ontology library system
E.g. RDFS, XMLS, DAML+OIL
Upper-level Ontologies

32. Organising Ontologies (7)
Current storage systems
Adaptation
Searching
Access through the Internet or World Wide Web
Offer Simple browsing only
SQL-based searching
Editing
Simple editing function
Reasoning
Very simple reasoning function
E.g. WebOnto(rule-based reasoning), Ontolingua (ontology testing), SHOE(ontology revision)

33. Organising Ontologies (8)
Requirement for a Storage System
Other requirements
Ontology scalability
Maintenance facility
Explicit documentation

34. Translating between Ontologies
XSLT stands for Extensible Stylesheet Language Transformations
XSLT is used to transform XML documents into other kinds of documents--usually, but not necessarily, XHTML
XSLT uses two input files:
The XML document containing the actual data
The XSL document containing both the “framework” in which to insert the data, and XSLT commands to do so

35. Very simple example File data.xml: File render.xsl:
<?xml version="1.0"?> <?xml-stylesheet type="text/xsl" href="render.xsl"?> <message>Howdy!</message>
File render.xsl:
<?xml version="1.0"?> <xsl:stylesheet version="1.0” xmlns:xsl=" <!-- one rule, to transform the input root (/) --> <xsl:template match="/"> <html><body> <h1><xsl:value-of select="message"/></h1> </body></html> </xsl:template> </xsl:stylesheet>

36. The .xsl file An XSLT document has the .xsl extension
The XSLT document begins with:
<?xml version="1.0"?>
<xsl:stylesheet version="1.0" xmlns:xsl=" XSL/Transform">
Contains one or more templates, such as:
<xsl:template match="/"> ... </xsl:template>
And ends with:
</xsl:stylesheet>

37. Finding the message text
The template <xsl:template match="/"> says to select the entire file
You can think of this as selecting the root node of the XML tree
Inside this template,
<xsl:value-of select="message"/> selects the message child
Alternative Xpath expressions that would also work:
./message
/message/text() (text() is an XPath function)
./message/text()

38. Putting it together
The XSL was: <xsl:template match="/"> <html><body> <h1><xsl:value-of select="message"/></h1> </body></html> </xsl:template>
The <xsl:template match="/"> chooses the root
The <html><body> <h1> is written to the output file
The contents of message is written to the output file
The </h1> </body></html> is written to the output file
The resultant file looks like: <html><body> <h1>Howdy!</h1> </body></html>

39. How XSLT works
The XML text document is read in and stored as a tree of nodes
The <xsl:template match="/"> template is used to select the entire tree
The rules within the template are applied to the matching nodes, thus changing the structure of the XML tree
Unmatched parts of the XML tree are not changed
After the template is applied, the tree is written out again as a text document

40. Where XSLT can be used
With an appropriate program, such as Xerces, XSLT can be used to read and write files
A server can use XSLT to change XML files into HTML files before sending them to the client
A modern browser can use XSLT to change XML into HTML on the client side
Most users seldom update their browsers
If you want “everyone” to see your pages, do any XSL processing on the server side
Otherwise, think about what best fits your situation

41. xsl:value-of
<xsl:value-of select="XPath expression"/> selects the contents of an element and adds it to the output stream
The select attribute is required
Example (from an earlier slide):
<h1> <xsl:value-of select="message"/> </h1>

42. xsl:for-each xsl:for-each is a kind of loop statement
The syntax is <xsl:for-each select="XPath expression"> Text to insert and rules to apply </xsl:for-each>
Example: to select every book element irrespective of where it is (//book) and make an unordered list (<ul>) of their titles (title), use: <ul> <xsl:for-each select="//book"> <li> <xsl:value-of select="title"/> </li> </xsl:for-each> </ul>
// means any ancestor
/ means just the parent

43. Filtering output
You can filter (restrict) output by adding a criterion to the select attribute’s value:
<ul> <xsl:for-each select="//book"> <li> <xsl:value-of select="title[../author=‘Dan Brown']"/> </li> </xsl:for-each> </ul>

44. Filter details
Here is the filter we just used: <xsl:value-of select="title[../author=‘Dan Brown'"]/>
author is a sibling of title, so from title we have to go up to its parent, book, then back down to author
This filter requires a quote within a quote, so we need both single quotes and double quotes
Legal filter operators are: = != < >

45. But it doesn’t work right!
Here’s what we did: <xsl:for-each select="//book"> <li> <xsl:value-of select="title[../author=‘Dan Brown']"/> </li> </xsl:for-each>
This will output <li> and </li> for every book, so we will get empty bullets for authors other than Dan Brown
There is no obvious way to solve this with just xsl:value-of

46. xsl:if
xsl:if allows us to include content if a given condition (in the test attribute) is true
Example: <xsl:for-each select="//book"> <xsl:if test="author=‘Dan Brown'"> <li> <xsl:value-of select="title"/> </li> </xsl:if> </xsl:for-each>
This does work correctly!

47. xsl:choose
The xsl:choose ... xsl:when ... xsl:otherwise construct is XML’s equivalent of Java’s switch ... case ... default statement
The syntax is: <xsl:choose> <xsl:when test="some condition"> some code </xsl:when> <xsl:otherwise> some code </xsl:otherwise> </xsl:choose>

48. xsl:sort You can place an xsl:sort inside an xsl:for-each
The attribute of the sort tells what field to sort on
Example: <ul> <xsl:for-each select="//book"> <xsl:sort select="author"/> <li> <xsl:value-of select="title"/> by <xsl:value-of select="author"> </li> </xsl:for-each> </ul>
This example creates a list of titles and authors, sorted by author

49. xsl:text
<xsl:text>...</xsl:text> helps deal space related problems:
XSL isn’t very careful with whitespace in the document
This doesn’t matter much for HTML, which collapses all whitespace anyway (though the HTML source may look ugly)
<xsl:text> gives you much better control over whitespace; it acts like the <pre> element in HTML

50. Creating tags from XML data
Suppose the XML contains
<name>Dr. Dave's Home Page</name> <url>
And you want to turn this into <a href=" Dr. Dave's Home Page</a>
We need additional tools to do this

51. Creating tags
Suppose the XML contains <name>Dr. Dave's Home Page</name> <url>
<xsl:attribute name="..."> adds the named attribute to the enclosing tag
The value of the attribute is the content of this tag
Example:
<a> <xsl:attribute name="href"> <xsl:value-of select="url"/> </xsl:attribute> <xsl:value-of select="name"/> </a>
Result: <a href=" Dr. Dave's Home Page</a>

52. Modularization
Modularization--breaking up a complex program into simpler parts--is an important programming tool
In programming languages modularization is often done with functions or methods
In XSL we can do something similar with xsl:apply-templates
For example, suppose we have a DTD for book with parts titlePage, tableOfContents, chapter, and index
We can create separate templates for each of these parts

53. Book example
<xsl:template match="/"> <html> <body> <xsl:apply-templates/> </body> </html> </xsl:template>
<xsl:template match="tableOfContents"> <h1>Table of Contents</h1> <xsl:apply-templates select="chapterNumber"/> <xsl:apply-templates select="chapterName"/> <xsl:apply-templates select="pageNumber"/> </xsl:template>
Etc.

54. xsl:apply-templates
The <xsl:apply-templates> element applies a template rule to the current element or to the current element’s child nodes
If we add a select attribute, it applies the template rule only to the child that matches
If we have multiple <xsl:apply-templates> elements with select attributes, the child nodes are processed in the same order as the <xsl:apply-templates> elements

55. Applying templates to children
<book> <title>XML</title> <author>Gregory Brill</author> </book>
<xsl:template match="/"> <html> <head></head> <body> <b><xsl:value-of select="/book/title"/></b> <xsl:apply-templates select="/book/author"/> </body> </html> </xsl:template> <xsl:template match="/book/author"> by <i><xsl:value-of select="."/></i> </xsl:template>
With this line: XML by Gregory Brill
Without this line: XML

56. Calling named templates
You can name a template, then call it, similar to the way you would call a method in Java
The named template: <xsl:template name="myTemplateName"> body of template </xsl:template>
A call to the template: <xsl:call-template name="myTemplateName"/>
Or: <xsl:call-template name="myTemplateName"> parameters </xsl:call-template>

57. Templates with parameters
Parameters, if present, are included in the content of xsl:template, but are the only content of xsl:call-template
Example call: <xsl:call-template name="doOneType"> <xsl:with-param name="header" select="'Lectures'"/> <xsl:with-param name="nodes" select="//lecture"/> </xsl:call-template>
Single quotes inside double
quotes make this a string
This parameter is a
typical XPath expression

58. Thoughts on XSL
XSL is a programming language--and not a particularly simple one
These slides have been an introduction to XSL and XSLT--there’s a lot more of it we haven’t covered
As with any programming, it’s a good idea to start simple and build it up incrementally: “Write a little, test a little”
This is especially a good idea for XSLT, because you don’t get a lot of feedback about what went wrong

59. FYI There exists also Formatting Objects Processors
Task is to transform XML into other formats
Not just other XML type documents
But also PDF, PS, PNG, etc

60. Automatic Ontologies Maintenance
Use IE + WWW search to identify them
New Domain Word identification
Biocomputation
Abbreviations (OWL)
Aliases (BSE also known as “Mad Cow Disease”)
Spelling Errors
Anesthesia > Anaesthesia
Orthographic Variances
Audio/Visual > Audio-visual
Morphological Variances
CD ROM > CD ROMs

61. Related Concepts A term G is related to term T if
T and G share some words
Ballistic missile -- medium-range ballistic missile
T and G often appear together in same sentences
Select a set of semantically related terms with higher domain specificity

62. Relation Extraction Structurally Suggested ISA Relation
Position defense. The type of defense in which ….
Hyperspectral imagery. A term used to describe the imagery derived from ..
Lexically Suggested ISA Relation
guided rocket-powered delivery vehicle ---ISA--- delivery vehicle

63. Automatic Ontology Construction (1)
Agroterrorism. Terrorist attacks aimed at reducing the food supply by destroying crops using natural pests such as the potato beetle, animal diseases such as hoof and mouth disease and anthrax, molds and other plant diseases.

64. Automatic Ontology Construction (2)
Food Supply
Potato Beatle
Attack
IS A
reduce
Crop
Hoof
IS A
IS A
destroy
Terrorist Attack
Natural Pest
Mouth Disease
IS A
IS A
use
Animal Disease
Anthrax
Agroterrorism
IS A
use
Plant Disease
Mold
IS A
use

65. Conclusion Ontology maintenance is a critical problem Need support
Tools help
Theories help
Text analysis helps

66. Questions?