Adaptive and Metaadaptive Hypermedia Applications

1
Adaptive and Meta-adaptive Hypermedia Applications

• Patricia Seefelder de Assis1,2
• Daniel Schwabe1
• 1 PUC-RJ
• 2 IPRJ / UERJ

2
Adaptive Hypermedia Systems (AHSs)

• Hypermedia Systems
• Structured set of nodes (information items) that
  can be navigated
• Example WWW
• Good Hypermedia Applications
• Rich link structure
• Freedom to navigate
• High degree of user interaction
• Freedom vs. usability
• Adaptation
• Versus one size fits all
• To optimize the use of Hypermedia Systems
• Adaptive Hypermedia Systems
• Tailor the application to their users.

3
Adaptive Hypermedia Systems

• Change Presentation Navigation Content
• According to User profile, Navigation history,
  Application usage, Runtime environment
• Useful for Individuals with different goals,
  knowledge and background Large hyperspaces
  Usage in diverse environments

4
First Time
Second Time
http//wwwis.win.tue.nl/2L690/
AHA! Adaptive Hypermedia Architecture
5
AHSs Some techniques

• Conditional inclusion of fragments
• Direct navigation guidance
• Link annotation
• Link hiding
• Link sorting

6
Conditional inclusion of fragments
7
AHSs Some techniques

• Conditional inclusion of fragments
• Direct navigation guidance
• Link annotation
• Link hiding
• Link sorting

8
AHSs Some techniques

• Conditional inclusion of fragments
• Direct navigation guidance
• Link annotation
• Link hiding
• Link sorting

9
AHSs Some techniques

• Conditional inclusion of fragments
• Direct navigation guidance
• Link annotation
• Link hiding
• Link sorting

10
AHSs main application areas

• Educational hypermedia
• On-line information systems
• Hypermedia for Information Retrieval

Brusilovsky, P. Adaptive Hypermedia. In User
Modeling and User Adapted Interaction, 11
87-110, 2001
11
Educational Hypermedia

• Interests in providing distance education over
  the Web has been a strong driving force for
  research
• The hyperspace represents the learning material
  on a particular topic
• Usually, the goal of the user is to learn most or
  all of the material (can also be to review a
  concept)
• Typically, the goal of the system is to provide
  systematic introduction to information
• Knowledge of different users can vary greatly,
  and grows at different rates
• Navigational help needed

12
NetCoach
Goal Learn all
Goal Learn Topic
http//netcoach.orbis.de/NR2000
13
NetCoach
Goal Learn all
Goal Learn Topic
http//netcoach.orbis.de/NR2000
14
Online information systems

• Heterogeneous group that includes electronic
  encyclopedias, information kiosks, virtual
  museums, e-commerce systems (among others)
• Each node of the hyperspace typically represents
  one concept of the subject and contains several
  pages of information.
• Usually, the goal is to provide reference access
  to information (e.g. on-line documentation)
• Users with different knowledge and background
  need different information about a concept at
  different levels of detail.
• Navigation help needed to find relevant pieces of
  information in order to achieve their goals
• Ecommerce peculiarity browsing of the
  hyperspace is a byproduct of the major activity
  (the better these systems work, the less
  browsing should be required)

15
http//www-mmt.inf.tu-dresden.de/english/projekte/
TELLIM/
16
Hypermedia for information retrieval

• Significantly influenced by the Web
• The hyperspace is usually very large - links
  between documents are not created "by hand" but
  are calculated by the system.
• The challenge is to to support the user's
  retrieval activity in an unrestricted Web
  hyperspace.
• Two large groups
• search-oriented systems goal is to create a list
  of links to documents that satisfy the user's
  current information request
• browsing-oriented systems support their users in
  the process of search-driven browsing through
  standard adaptive navigation support technologies

17
http//www.cs.cmu.edu/webwatcher/
18
AHSs usual models

• Domain Model (DM)
• Structures the content
• Concepts relationships
• User Model (UM)
• Goals, preferences, knowledge
• Adaptation Model (AM)

19
Meta-adaptive Hypermedia Systems

• Adaptation can be applied to the Adaptation
  Mechanism according to given parameters
• Meta-adaptation two levels
• Adaptation techniques (they may work most
  efficiently in different contexts)
• Example More restrictive techniques for novice
  users
• Schema adaptation
• Example Inclusion of new inferred relationships
  in the UM

20
A General Meta-model for AHS

• To answer the questions
• What is being adapted?
• What is the adaptation based on?
• When?
• How is the adaptation achieved?
• Where?

21
What

• Structure of hypermedia
• Adaptive presentation (Content-level)
• Adaptive navigation support (Link-level)
• Main aspects of hypermedia applications
• Adaptive content
• Adaptive navigation
• Adaptive presentation
• Independence of main aspects
• Adaptive content
• Adaptive navigation
• Content structure
• Hyperspace topology
• Adaptive presentation
• Content interface
• Navigation support interface

Distinction of adaptation at presentation-level
changes to the layout that do not affect the
content.

• To characterize more precisely
• Distinction between adaptation of content itself
  and of structure of content
• Distinction between changes on content layout
  from changes on link layout.

22
Description
What is adapted
The actual contents colloquial text for
non-experts vs. technical text for experts
simple text for beginners vs. detailed text for
advanced users
Content
Node-level(the particular way in which concepts
are put together for navigation
purposes) including introduction for beginners/
skipping it for advanced users including safety
warning the first time a critical content is
browsed / not including it in subsequent visits.
Content structure
Navigation
Link-level (anchor changes and index changes
modify the navigation path) including a link to
solution in a problem used as an example / not
including the link in a problem used as a test.
Hyperspace topology
Fragment level Using fonts of different size
according to users' age highlighting certain
types of information for emphasis
Content interface
Presentation
Anchor level using drop-down menu / using
explicit lists of anchors using textual anchors
/ iconic anchors
Navigation support interface
23
Based on what

• Structure of content
• User profile and behavior
• Utilization

24
Explanation and example
Based on what
The way the content is structured in the DM may
affect the adaptation. a (hierarchically)
composed concept may be considered known by the
user only when all its elementary concepts are
known.
Domain Model
User preferences user characteristics user
role experienced or novice user prefers texts or
images preferred language whether the user is a
student or a teacher / is identified or
anonymous knowledge about domain goals
User Profile
Under which conditions the system is used.
personal computer vs. cell phone physical
location access bandwidth.
Context
Navigation path while browsing. pages previously
accessed.
Navigation History
Actions taken (interactions) while browsing.
articles read / printed how many times video
saw images enlarged.
Browsing Behavior History
How the functionality of the application is used.
results of tests previously bought items.
Applic.Functio-nality Usage
from spot to episodic to continuous time.
adaptation made at login time / after a user
configuration action system revision and
adaptation after each navigation step.
Granularity of time
25
How

• Adaptation Function

26
A General Meta-model for AHS

• Usual Models
• DM UM AM
• Clear separation
• Plus
• Navigation Model
• Presentation Model
• And
• Integration Model
• User Context Model

27
(No Transcript)
28
Adaptation Functionality

• What is being adapted (outputs)
• What is the adaptation based on (inputs)
• How is the adaptation achieved (Adaptation
  Function)

29
(No Transcript)
30
Comparison of AHSs

• Systems modeled and compared based on each
  meta-model
• Main adaptation questions analyzed

31
(No Transcript)
32
Meta-model summary

• Its difficult to developers to determine the
  effective adaptation techniques
• General meta-model
• Facilitates the adaptation design and evaluation
• Provides for meta-adaptation
• Capable of expressing different adaptation
  techniques
• Instances input to meta-adaptation rules
• Meta-adaptation rules act on data and on the
  models themselves
• BUT
• Conceptual, navigational and implementation
  aspects are mixed-up
• UML notation is not very expressive

33
Semantic Model

• Use of the Semantic Web language to specify
• Adaptation function (how)
• Input parameters (based on what)
• Output parameters (what)
• While maintaining the independence between
• Conceptual aspects
• Navigational aspects
• Implementation aspects.

34
Semantic Model ASHDM

• SHDM uses ontologies
• Conceptual ontology mapped into navigational
  ontology
• Abstract Interface Concrete interface
• ASHDM SHDM adaptation
• Conceptual Model any
• Navigation Model and Interface Model
• Acquire adaptive features
• User (Context) Model and Adaptation Model
• Included
• Adaptation architecture.

35
ASHDM Models

• Domain gt Conceptual
• Navigation
• Presentation gt Interface
• Integration
• Adaptation Context Model
• User Model Context Model
• Adaptation

36
Conceptual Model

• Any ontology defined for the Semantic Web
• Real World Representation
• Concepts relationships between them
• Application-independent
• Open domain
• integration through navigational views
• Not adapted itself
• Specific views through Navigation Model

37
Conceptual Model
Artist Iron Maiden CD The Number Of The Beast
(1982) Tracks Invaders (323)
IronMaiden rdftype cdsArtist cdsnme
Iron Maiden cdsrecords
TheNumberOfTheBeast . TheNumberOfTheBeast
rdftype cdsCD cdstitleThe Number Of The
Beast cdsyear 1982 cdscover
lthttp//www.ironmaiden.com/.../numberofthebeast.jp
ggt cdstrack rdftype cdsTrack
cdsnumber 1 cdsduration 203
cdsrecordOf invaders invaders
rdftype cdsMúsica cdstitle Invaders
.
38
Navigation Model

• Views over conceptual objects
• Based on
• Users tasks
• Users profile
• Mapping
• Integration Model
• Ontology-based vocabulary
• Classes
• Links
• Contexts
• Access structures

Attributes values
Nodes Navigation
39
Navigational ontology instance

• IronMaiden
• rdftype nSchemaArtista
• rdftype nSchemaNode
• nSchemanome Iron Maiden
• nSchemagrava TheNumberOfTheBeast .
• TheNumberOfTheBeast
• rdftype nSchemaCD
• rdftype nSchemaNode
• nSchematitle The Number Of The Beast
• nSchemayear 1982
• nSchemacover lthttp//www.ironmaiden.com/...
  /numberofthebeast.jpggt
• nSchemamusics
• nSchemanumber 1
• nSchematime 203
• nSchematitle Invaders
• ...

40
What Navigation

• Content Structure
• Example

41
What Navigation

• Topology of Hyperspace
• Example
• Link for an explanation text or for an example,
  according to users learning style

42
What Navigation

• Other Examples
• A conceptual attribute photo is not mapped in
  the navigational class, for a small display
• Learning style used to select the context topic
  by chapter or topic by exercise
• Button Next leads to chapter_1_extended or to
  chapter_2, depending on users knowledge
  status
• Link to an answer only created if the exercise is
  used as an example, according to users background

43
What Content

• The value of some attributes defines the content
  of the node
• Example

44
Interface Model

• Abstract Widget Ontology
• Interaction between user and system
• Concrete Widget Ontology
• Activators (Button Link)
• Exhibitors (Image Label)
• Capturers (CheckBox ComboBox RadioButton
  LinkElement)
• Mapping Abstract into Concrete
• CSS

45
Presentation Adaptation - Mapping

• ElementExhibitor
• Image
• Label
• SingleChoice
• Combobox / Action / Target
• RadioButton / Action / Target
• Link
• DiscretGroup
• Like above
• CheckBox / Action

Schematic instructions or textual ones
Defined according to the number of options and if
one can navigate to them (example read or print)
With CheckBox it is possible to select multiple
text to print or to exhibit
46
Presentation Adaptation - Layout

• Change CSS
• Change CSS properties (e.g., font color, weight,
  size)

47
Adaptation Context Model
DM

ACM

UM

UR

Adaptation Layer

NM

IM

ACM (Adaptation Context Model)

UM (User Model)

DM (Domain Model)

UR (User Representation)

NM (Navigation Model)

IM (Interface Model)


48
Adaptation Context Model
history
ID, name, cognitive_style, presentation_ preferenc
es
DM

ACM

UM

UR

display
ID, name, address, birthday, favorites
Age, interest_ categories, history, display
Adaptation Layer

NM

IM

ACM (Adaptation Context Model)

UM (User Model)

DM (Domain Model)

UR (User Representation)

NM (Navigation Model)

IM (Interface Model)


49
User Model

• UM independent from applications
• Central Knowledge Bases
• Sharing (similar domains)
• Reuse
• Task division (construction and maintenance
  onerous)
• Initialization vs. Accuracy
• Consistency for users
• Refinement of the model
• Known semantics

50
User Model Schema
51
Example e-commerce

• Customer view
• Static info like demographic and personality data
  used to infer preferences about products
• Dynamic data like kind of access and device being
  used may define which kind of information to show
  (textual / graphic)
• Overlay Model could include data from other
  stores (privacy and other questions ignored) to
  inform products already bought
• Domain-dependent data reading preferences for a
  bookstore sports skills for a sport one
• Goal own use vs. offer as a gift
• Supplier view other attributes need to be
  considered

52
Example e-learning

• Customer view
• Static and context info to adapt the presentation
• Preferences about subject cannot be considered!
  (Learner cannot choose not to learn physics)
• Overlay Model usually, what is already known
• Domain-dependent data learning style
• Parameters used to retrieve data from UM user
  role goals to achieve tasks to do.

53
E-learning peculiarities

• There is always an user and this user is always
  central to the application
• E-commerce buy to others
• Touristy kiosk passer-by access
• The Domain Model itself is an ontology relevant
  to to the adaptation process
• Info about product or about museum artifacts not
  as user-related as concepts to be learned
• Domain ontology vs. Task ontology
• UM task ontology

54
Adaptation Model
55
Execution Flow (1)
56
Execution Flow (2)
57
Example

• Professors listed alphabetically
• User click on a name (Schwabe) and triggers the
  adaptation process
• The instance of a navigational ontology for
  Schwabe is obtained through an adaptive mapping
  over the conceptual ontology, defining what will
  be presented and how

58

• IF the user is interested in personal data about
  Schwabe, the navigational schema select will be
  agenda-like (attributes title, e-mail, phone,
  homepage)
• OTHERWISE, it will be portfolio-like
  (attribute publications)
• The publications may be exhibited as one abstract
  per page or as titles-anchors that links to
  abstracts
• The Abstract Interface is instantiated

59
Exemplo
60
Meta-adaptation

• Adapted selection of the Adaptation Model,
  including the selection of the Execution Model
• change the adaptation strategy if the user
  browses repeatedly the same concept
• Adaptation of models
• according to the user profile, the exhibition of
  a security-alert the first time the user accesses
  a concept, may be implemented by
  presentation-adaptation rules (a heavier font the
  first time) or by content-adaptation rules
  (exhibition if the alert only the first time) or
  by not considering adaptation rules (exhibition
  always)
• the choice about what to include in the User
  Model made according to users goals

61
Implementation Architecture

• Model-driven implementation architecture
• Direct manipulation of
• its metamodels
• generated models
• relations between these models
• good for meta-adaptation
• HyperDE ASHDM

62
Implementation Architecture(HyperDE) - Overview
63
HyperDE Metamodel ASHDM Adaptation Model
64
Execution Flow (HyperDE)

• Control Layer
• Selection of anchor
• Index / Element
• Rules
• Model Layer
• SHDM primitives
• Parameters
• Rules
• Adapted Application Navigation Model
• Views gt Adapted Page

65
Example 1 - description

• Metaclass
• Publication
• Attributes
• Abstract
• Short_abstract
• Extended_abstract
• Rule
• Rule1 associated with Abstract

66
Example 1- DSL
Classes Publication attrs
abstract rule1 hook pre
value if AdaptationContext.c
urrent.user.expertise novice
attrName short_abstract
else attrName
extended_abstract end
67
Example 1- DSL
Classes Publication attrs
abstract rule1
value if AdaptationContext.curren
t.user.expertise novice
else
end
hook post
val self.short_abstract
val self.extended_abstract
68
Another (DSL) Example
69
ASHDM vs. AHA!
70
Conclusion

• The theory behind a general architecture for
  adaptation and meta-adaptation of Hypermedia
  Systems was presented
• The presented model is extremely flexible, maybe
  too much flexible
• Necessity of a case study
• To prove the value of the model, while adjusting
  it