and Customisable Ontology Evolution

snufflevoicelessInternet and Web Development

Oct 22, 2013 (3 years and 5 months ago)

76 views

Tuesday , May 24, 2011

Ontology
-
based Content Change Management
and Customisable Ontology Evolution

Muhammad Javed

PhD Researcher (Dublin City University)

SZABIST, Islamabad



2
/23

Agenda

Introduction

Motivation

Research Problem

Research Work

Q/A Session



3
/23

Software System

SW. Components
s

Tasks



Archiving


Installing


Configuring

Activities


Restoring Items


Archiving Folders

SW. Features


GUI Elements

-

like Structure

Software
System

SW. Feature

Files

Database

Activities

DocBook

SW. Comp.

Commands

Procedures

H&Q Struc.

hasProcedure

4
/23

Introduction

With the rapid growth of
semantic web
,
ontology
-
driven
modelling

of multi
-
formatted
content

is vital for a wide
range of enterprise
knowledge management systems
.


By saying content, we mean digital information

Unstructured/Semi
-
structured Text (html/xml documents)

source code (e.g. java script),

executable element (e.g. applet),

software elements (e.g. GUI feature)

etc.





Textual representation of


software system components


in form of such as web files

(html/xml format)


5
/23

Introduction
(contd.)

Ontologies can be used to
represent

and

access

the
content semantically.


The dynamic nature of content requires domain ontologies
to evolve over time.

Introduction/removal of entities

change in the structures (generalization/ specialization)

change in the meanings of entities.


Domain ontologies must be
consistent

with the content for
their effective usage.





6
/23

Motivation

Potential of ontologies

for replacing the traditional content representation

semantic access to the content.


A single change in domain ontology or in content

may lead to a number of inconsistencies

some knowledge may get inaccessible

some knowledge may lose its semantics


Currently developed models

do not
deal with such
management and consistency issues.




7
/23

Motivation
(contd.)

There is no single ontology change management system
which is widely accepted and provides different levels of
change operators based on

granularity

domain specificity

Abstraction


Ontology change representation procedures are not
sufficient to represent how ontology evolves over time.



8
/23

<Installing>

<Integrated Search>

<Planning>

<ArchiveExplorer>

<Configuring>

<NetworkAccessControlEnforcer>

<Archiving>

<Introducing>

<ImplementationGuide>



9
/23

Research Problem

How to content changes semantically
using domain ontologies?





manage

10
/23

Research Questions




-

content representation using the domain ontologies.


-

manage
content changes to keep them consistent with domain ontologies.





-

change operators based on granularity and domain
-
specificity.


-

representation of
commonly used ontology change patterns.






-

ontology change representation


-

identification of change patterns from the elementary level change representation.

Part 1:

Ontology
-
based Content Modelling

Part 2:

Ontology Change Customisation

Part 3:

Ontology Change Representation



11
/23

Ontology Change
Customisation

12
/23


We studied the evolution of ontologies empirically
in order to

Investigate the relationships between
generic

and
domain
-
specific

changes.

Determine common patterns of change.


Selection of Domain:

Database System Ontology

University Administration Ontology

Part 2 : Ontology Change Customisation



13
/23

Results of Empirical Study:

Database Ontology:

Changes are identified by taking different perspectives
into account.

Teaching

Publishing

Industry

University Administration Ontology:

Patterns are identified by observing the practical
changes in the domain

faculty management

joining/departure of staff

student registration

etc.



14
/23

Framework of Compositional Operators and Change Patterns [1]:

-

level one: elementary changes which are atomic tasks.

-

level two: aggregated changes to represent composite, complex tasks.

-

level three: domain specific change patterns.

-

level four: abstraction of the domain specific change patterns.


15
/23

An Example:

Adding a New Chapter:


Change Operations 1, 3 and 9


Deleting a Chapter:


Change Operations 5 and 10


Splitting a Chapter:


Change Operations 6, 9 and10



-

Specific to the domain and the requirements of the ontology user, with out going
to the details of the first and the second levels, he can choose his own patterns
and execute change based on the patterns defined before.

16
/23

Ontology Change
Representation

17
/23

Research work being done in past represent the ontology changes at the elementary
level, which does not suffice their utility.


The ontology engineer is unable to understand

why he did such changes,

Whether it’s a basic level change or a part of composite change

what the impact of such changes on the ontology was.


3.1 Layered Change Log

We have introduced
Pattern Change Log
[2]

and proposed a
layered change log
model
, containing two different levels of granularity, i.e.

Basic Change Log (BCL)

Pattern Change Log (PCL)

Part 3 : Ontology Change Representation



18
/23

Change Representation

Five Aspects:


1)
WHO

performed the Change (User)

2)
WHEN

this change is applied (Timestamp)

3)
HOW

I find the change in change list (Change Id)

4)
WHAT

is the change (Operation & Element)

5)
WHERE

the change is applied in the ontology
-
context (Parameters)

Meta Data

Change Data

Single Ontology Change

(John, PhD_Student)

1299772286625

Javed

Thu Mar 10 15:51:26 GMT 2011

Add classAssertion

19
/23

Generic Attribute Type Graph

Graph Node

Attribute Node


metadata.

Attribute Node


change data.

Node Attribute Edge

Edge Attribute Edge

Ontology Change Formalisation:

20
/23

Attributed Graph Typed Over ATG

t(Add) =


Atomic Operation

t(
classAssertion
) =

Individual Axiom

t(Phd_Student) =

Entity (Concept)

t(John) =


Entity (Individual)

Graph Edge

Linear Directed Graph

Node
5

Node
6

Node
7

Ontology Change Formalisation:

21
/23

Empirical Analysis of Change Log Graph

We observed bundles of change operations that are

frequent

in the change log and

represented as a
chain of atomic changes (level 1)

.


We identified such change operation sequences and
used them as a reference to discover
usage
-
driven
change patterns
.


The
algorithms

to discover such change patterns has
been written and will be published soon.

22
/23

Research Publications:

Research Papers
:


1.
M. Javed, Y. Abgaz, and C. Pahl


A Pattern
-
based Framework of Change Operators for Ontology Evolution.


In Proc: The 4th International Workshop on Ontology Content (OntoContent), volume 5872 of Lecture Notes in Computer Science,
pag
es
544

553. Springer, November 2009.

2.
M. Javed, Y. Abgaz, and C. Pahl


A Layered Framework for Pattern
-
Based Ontology Evolution


3rd International Workshop on Ontology
-
Driven Information System Engineering (ODISE), 20th June 2011, London, UK.


3.
C. Pahl, M. Javed and Y. Abgaz

Utilising Ontology
-
based Modelling for Learning Content Management.


In Proc: ED
-
MEDIA 2010
-
World Conf. on Educational Multimedia, Hypermedia & Telecommunications, 29
th

June


2
nd

July 2010,
Toronto, Canada.

4.
Y. Abgaz, M. Javed and C. Pahl


Empirical Analysis of Impacts of Instance
-
Driven Changes in Ontologies


In Proc: The 6th International Workshop on Ontology Content (OntoContent), 25
th



29
th

October 2010, Crete, Greece.

5.
M. Javed, Y. Abgaz, and C. Pahl


Ontology
-
based Domain Modelling for Consistent Content Change Management .
I


In Proc: The
International Conference on Ontological and Semantic Engineering (ICOSE), 24
th

-

26
th

November

2010, Venice, Italy.


Posters
:

1.
M. Javed, Y.M. Abgaz and C. Pahl


Manipulation of Evolving Ontology
-
based Content Systems


Public Localisation Showcase; In conjunction with Innovation Dublin, 10 November 2010, Microsoft Ireland, South County Dublin

Business Park, Ireland.


2.
M. Javed and C. Pahl


A Change Operator Calculus for Customizable Ontology Evolution.


Innovation Showcase Workshop, October 2009, Dublin, Ireland

3.
M. Javed and C. Pahl

A Fine
-
granular Semantic Annotation for Content Change Management.

Annual Scientific meeting of Centre for Next Generation Localisation (CNGL), 27 April 2010, University of Limerick, Ireland.

4.
M. Javed and C. Pahl


Consistent Change Management for Semantically Annotated Content.


Annual Scientific meeting of Centre for Next Generation Localisation (CNGL), 28 April 2010, University of Limerick, Ireland.


www.computing.dcu.ie/~mjaved



23
/23

Thank You !



If

greatness

of

purpose,

smallness

of

means

and

outstanding

results

are

the

three

criteria

of

human

genius,

who

could

dare

to

compare

any

great

man

in

modern

history

with

Muhammad

(PBUH)
.


Alphonse de La Martaine