View
46
Download
0
Category
Preview:
Citation preview
Content and Service Adaptation Management
By
Abdelhak Attou
Supervised by
Prof Klaus Moessner
Dr Francois Carrez
Overview
• Problem Domain
• Motivations and Objectives
• Contributions
– Adaptation Context Management
– Description Logics Based Rule Based Reasoning
• Implementation/Evaluation
• Conclusions
Terminology
• Adaptation
– Selection/generation/modification of content/services
for a delivery contexts upon user requests
• Delivery/Adaptation context
– device/network descriptions, user preferences and
characteristics, consumption environment description,
content and service metadata
• Content
– Text, image , video, audio or multimedia objects
• Services
– Delivery and presentation of content
Adaptation
• Static
– Generate different versions at authoring time
• E.g. BBC PC and Mobile versions
• Device centric
• Impossible to cover all delivery contexts
• Dynamic
– Generate the required version at request time
• Requires more computational resources
• User centric , offers personalization
• Important to gain and maintain customers
The Semantic Web Description Logics (DL) vs. Ontology vs. OWL
Ontology is a philosophical concept
– Study of concepts, being and existence
DL describes an ontology mathematically
– Using First Order Predicate Logic
– Powerful knowledge representation mechanism
OWL (Ontology Web Language)
– (XML + DL) based ontology language
– Latest version is OWL 2.0
Content Adaptation Domain
Complex
HeterogeneousDynamic/Volatile
Numerous Tools
Research Objectives
• Identify main problems and requirements of content and service
adaptation systems
• Evaluate available tools.
• Propose effective mechanisms to meet the identified
requirements.
• Investigate the use of Semantic Web technologies in content
and service adaptation
Adaptation Systems :
Problems and Requirements
• Problem: Heterogeneity
– Requires : Interoperability
• Problem: Dynamicity
– Requires : Extensibility, Maintainability
• Problem: Complexity
– Requires: Modularity, Management
Adaptation Management levels
ADMEADME
ADMEADME
ADMEADME
User
Context
Device
Context
ADME
Context
Content
Context
ADMEADME
General
ADTE
Specific
ADTESpecific
ADTESpecific
ADTE
Delivery
Channel
Delivery
Channel
Delivery
Channel
Content
Types
Content
Types
Content
TypesContent Selection
Context Management
Adaptation Decision Taking
Adaptation Execution Management
Delivery Management
This figure will need to be revised, and should be part of the state of the art (what do we
mean by management, can it be used in all stages ?)
Context Management
Adaptation Decision Taking
Contributions
• Extensible, maintainable, modular and
interoperable adaptation context processing
• Novel approach to extend the use of Description
Logics to rule based adaptation decision taking
Proposed Solution
Adaptation
Context Manager
Adaptation Manager
Adaptation Execution
Reasoning
ManagerAdaptation
Decision Engine
AM Gateway
Adaptation Context Management
Adaptation Context Issues
• Context is of paramount importance
• New context parameters / adaptation scenarios
need to be considered for personalization
• Need to manage the different context formats
– Of different syntax and vocabulary and semantics
• Issues that need to be addressed
– Hardcoded context formatting to internal model poses
extensibility and maintainability problems
– No focus on interoperability and extensibility
Proposed Solution
Hig
h-L
ev
el C
on
tex
t
Inte
rfac
e
As
sim
ilatio
n C
on
tex
t Inte
rfac
e
MPEG7/21
ContextHigh | Low
Level | Level
ADO
UAProf
Extracted
context
Descreasing complexity
Adaptation Context Management
Adaptation
Decision
Taking
Ad
ap
tatio
n
Co
nte
xt
Ad
ap
tatio
n C
on
text
Higher Level Context
T1
Tn
Context
Reasoner
ADO : an Adaptation Domain Ontology T : Context Transformer / Formatter
Proposed Solution
Hig
h-L
ev
el C
on
tex
t
Inte
rfac
e
As
sim
ilate
d C
on
tex
t Inte
rfac
e
MPEG7/21
ContextHigh | Low
Level | Level
ADO
UAProf
Extracted
context
Descreasing complexity
Adaptation
Decision
Taking
Ad
ap
tatio
n
Co
nte
xt
Ad
ap
tatio
n C
on
text
Higher Level Context
T1
Tn
Context
Reasoner
ADO : an Adaptation Domain Ontology T : Context Transformer / Formatter
Context Assimilation• Aims for
– Interoperability with existing
formats
– extensibility to new formats with
ease of maintenance
• Formal context modelling
– Description Logics + XML =
OWL
• Standard vocabulary
– MPEG-7/21
• Extensible formatting
– XSLT based
Proposed Solution
Hig
h-L
ev
el C
on
tex
t
Inte
rfac
e
As
sim
ilate
d C
on
tex
t Inte
rfac
e
MPEG7/21
ContextHigh | Low
Level | Level
ADO
UAProf
Extracted
context
Descreasing complexity
Adaptation Context Management
Adaptation
Decision
Taking
Ad
ap
tatio
n
Co
nte
xt
Ad
ap
tatio
n C
on
text
Higher Level Context
T1
Tn
Context
Reasoner
ADO : an Adaptation Domain Ontology T : Context Transformer / Formatter
Proposed Solution
Hig
h-L
ev
el C
on
tex
t
Inte
rfac
e
As
sim
ilate
d C
on
tex
t Inte
rfac
e
MPEG7/21
ContextHigh | Low
Level | Level
ADO
UAProf
Extracted
context
Descreasing complexity
Adaptation
Decision
Taking
Ad
ap
tatio
n
Co
nte
xt
Ad
ap
tatio
n C
on
text
Higher Level Context
T1
Tn
Context
Reasoner
Context Refinement • Define high level context concepts
– Act like adaptation rules
– Reduces decision taking complexity
– Reduces the impact of context
changes to the rest of the system
• Use latest features of OWL 2.0
– Data types reasoning
– Negative property assertions
Context Management
• The Adaptation Decision Interface Ontology
(ADIO)
– Internal context model
• Pre-reasoning Parameters/Functions
– Aims to enhance extensibility at minimal maintenance
– Aims to improve performance of context reasoning
Context Processing Stages
Context
Profile
Format
Context
Pre-
reasoning
Functions
Context
ProfileContext
Profile
Content \
Service
Extract
Context
Formatting
ToolFormatting
ToolFormatting
Tool
Refine
Context
Calc Pre-
Reasoning
Prameters
Adaptation
Request
ADIO
Refined ADIO
ADIO
ProcessInput / Output Input / Output
• Defines low level concepts based on
– MPEG-7- Multimedia Description Interface
– MPEG-21 – Digital Item Adaptation
• Defines high level concepts such as
•
•
•
ormat(mp4)supportedF ormat(mp4) requiredF mp4Limited
x Limitation bandwidth imitedBandwidthL
image)sModality(userPrefer edVideoLimit imitedBandwidthL
eryLimitedVideoDeliv
Adaptation Decision Interface Ontology
High Level concepts
• Example : defining BatteryLimitation based on
– Remaining battery
– Service battery requirements
–
• Variable x is assigned for each request :
– x = serviceBatteryRequirement
• Requires changing the ontology structure and
hence requires reclassification (not efficient)
• How extensible is context handling ?
xBattery remaining
• Assume we need to add
– user remaining battery preference ,
• E.g. User wants %20 of battery power remaining
–
– x = serviceBatteryRequ– userRemainingPref
– Requires changing the code that handles the ontology
– Not practical in such a dynamic domain
• Where user requirements and device/content
technologies change rapidly
Extending to new context parameter
xBattery remaining
Pre-Reasoning Functions/Parameters
Pre-reasoning parameters
• Avoid changing the ontology for each request
– Hence avoid reclassification and improve performance
• Extend to new context parameters with minimal
maintenance
Pre-reasoning Parameters
• Defining BatteryLimitation
– Remaining battery : r
– Service battery requirements : s
– User remaining battery preference : u
– A new data type property is created: batteryLimitation
•
• Pre-reasoning parameter: batteryLimitation
• Pre-reasoning function
– batteryLimitation = r - s – u
• Calculated for each request
• Does not requires changing the ontology
– Improved performance
0 mitation batteryLi
Managing Pre-reasoning Functions
• Represented as postfix expressions in XML
– Infix (a + b) – (c /d), Postfix b a + c d / -
– Postfix expressions are simpler to evaluate
– Extensibility and maintainability ensured by modelling
the postfix expressions in XML
Description Logics Rule Based Reasoning
Our Approach
• We use Description Logics and Ontology to
define adaptation rules
– Outside their use in state of the art (defining adaptation
concepts, e.g. Device, User, etc)
– To extend their expressiveness , for example to express:
Brother (A, B) , Father (A, C) Uncle (B,C)
• We implemented a Description Logics Rule Based
Reasoner for adaptation decision taking
Motivation
Semantic web is gaining momentum
– Description Logics is the cornerstone
– Continuous research on
• Expressiveness extensions
• Optimization techniques
Any platform that supports semantic web
– Will support a DL language and a DL reasoner
– Support of our approach would be straight forward
DL built in inference can be used to check
– Rule-base consistency
The DL based Rule Based Reasoner
Description Logics
Reasoner
Description Logics
based Rule Engine
Input : Context (ADIO)
Output : Adaptation Decisions
Rules Ontology
Pellet
Fact ++
Racer
The Approach
RuleBody RuleHead
The Approach : Structure
OWL:Thing
Rule
RuleBody
RuleHead
Equivalence
Sub-typing
Thing) : (OWLconcept op:
RuleHead RuleBody Rule
T
The Approach : Structure
OWL:Thing
Rule
RuleBody
RuleHead
lassanonymousC namedClass premise_n
_.... premise_2 premise_1
npremise
Premise_1
Adaptation
Context
Premise_1Premise_1
Premise_1{Equivalence
Sub-typing
The Approach : Structure
Body_n
Premise_1
Adaptation
Context
Premise_1Premise_1
Premise_1{
OWL:Thing
Rule
RuleBody
RuleHead
}_.... premise_2 {premise_1 Body_n npremise
Equivalence
Sub-typing
RuleBody Body_n
The Approach : Structure
Body_n
Rule_n
Premise_1
Adaptation
Context
Premise_1Premise_1
Premise_1{
OWL:Thing
Rule
RuleBody
RuleHead
Equivalence
Sub-typing
Body_n Rule_n Rule Rule_n
The Approach : Structure
Body_n
Head_n
Rule_n
Premise_1
Adaptation
Context
Premise_1Premise_1
Premise_1{
OWL:Thing
Rule
RuleBody
RuleHead
Equivalence
Sub-typing
Head_n Rule_n RuleHead Head_n
The Approach : Structure
Body_n
Head_n
Rule_n
Premise_1
Adaptation
Context
Premise_1Premise_1
Premise_1{
OWL:Thing
Rule
RuleBody
RuleHead
conclusion
Decisions
conclusionconclusion
conclusion_1
Equivalence
Sub-typing
lassanonymousC namedClass conclusion
}_.... _2conclusion n_1{conclusio Head_n
nconclusion
Implementation and Evaluation
Evaluation
• Evaluation in adaptation systems
– Evaluation by implementation and functional testing
• Demonstrate the feasibility of the proposed
solutions
– Performance testing
• Demonstrates the applicability of the proposed
solutions
– Demonstration use cases
• Evaluate certain features of the system
• Essential for this research to evaluate
extensibility, interoperability and maintainability.
Adaptation Manager Implementation
rule-based
ADTE
DL-KB
DL- resoning
Handler
Context
Extractor
Context
FormatterA.Operations
Handler
AdaptationContext
Manager
MP
EG
-7
(MD
S)
F
Reasoning Manager
DL
ResonerAdaptation Decision Engine
DIA-based
ADTE
DL rules
reasoner
Adaptation Gateway
Clie
nts
Ma
ng
erWeb services
Interface
CA interface
Rule-based ADTE interface
Adaptation Manager
Me
ss
ag
e
Tra
ns
lato
r
Content
Adaptor
Content & Service
Providers
Adaptation
clients
MP
EG
-21
(DIA
) F
The Content Adaptation Test Bed
Content Server
&
Adaptation Mechanisms
Server
WLAN
Client
Gateway
Adaptation
Management
Framework 1
Adaptation
Management
Framework 2Client
Results
• Scenarios 1
– MOV Video - 3GPP Video
• Device does not support MOV
• Battery limitation
• Resolution limitation
• Scenario 2
– MOV Video - MP3 Audio
• Device does not support video
• User driving
Performance Testing
Demonstration Use Cases Evaluation
Use case 1
• Supporting a new context profile
– E.g. User Agent Profile UAProf
Supporting UAProf
• Step 1
– Implement the XSLT formatter
ADIO UAProf
Supporting UAProf
• Step 2
– Add the formatter to the system via the formatters XML
file
Demonstration Use Cases Evaluation
Use case 2
• Extending to the service summary scenario
– Service summary may be required if
• User is not interested in videos longer than
certain length
• User is commuting/in their brake and want to
choose from different service to view later
• Saving resources
Extending to the Summary Scenario
ength maxVideoL- hvideoLengt esholdyLengthThrvideSummar
0} resholdryLengthThvideoSumma {
ty(brake)}userActivi { (train)} onuserLocati ({ required-summary
• Extending the context model
Conclusions
• Identified main problems and requirements for adaptation
management systems
– Extensibility, interoperability, maintainability and modularity
• Introduced an context adaptation management to ensure
these requirements are met
• Implemented the Adaptation Manager, part of the content
adaptation test bed
– Demonstration use cases scenarios shows that the proposed
mechanisms enhance extensibility maintainability and
interoperability
• Description Logics is becoming a dominant reasoning tool
– A Description Logics based adaptation decision taking approach is
proposed
Publications
• Patent
– Apparatus and Method for Performing Service Adaptation in Respect of a Mobile Computing
Device , USPO No. 12/235,455, application filed in September 2008
• Journals/Conferences
– A. Attou and K. Moessner, "Context-Aware Service Adaptation Management," presented at
Personal, Indoor and Mobile Radio Communications, 2007, PIMRC 2007, IEEE 18th
International Symposium on, 2007.
– A. Attou, D. Jie, D. I. Laurenson, and K. Moessner, "Performance Modelling and Evaluation of
an Adaptation Management System," presented at Performance Evaluation of Computer and
Telecommunication Systems, 2008, SPECTS 2008, International Symposium on, 2008.
– A. Attou, N. Li, and K. Moessner, "Context for Multimedia Services Adaptation” in 4th
International Mobile Multimedia Communications Conference. Oulu, Finland, 2008.
– S. Merat, A. Attou, and A. Hamid, "Content Adaptation: Requirements and Architecture," in
Proceedings of the 10th International Conference on Information Integration and Web-based
Applications /& Services. Linz, Austria: ACM, 2008.
– N. Li, A. Attou, and K. Moessner, "A MPEG-based Ontology for Ubiquitous Content/Service
Adaptation," in Context Awareness for Proactive Systems, Third International Workshop on,
Proceedings-ISBN 978-0-9556240-0-1, Guildford , UK, 2007.
– N. Li, A. Attou, S. De, and K. Moessner, "Device and Service Descriptions for Ontology-Based
Ubiquitous Multimedia Services," in Proceedings of the 6th International Conference on
Advances in Mobile Computing and Multimedia. Linz, Austria: ACM, 2008.
– A. Attou , N, Li and K. Moessner, "Service Adaptation Management," Multimedia , ACM
Transactions On, Under Review, 2009.
Questions ???
Recommended