ADAPT Major Design Dimensions for Educational Adaptive Hypermedia

1
ADAPT Major Design Dimensions for Educational
Adaptive Hypermedia

• Franca Garzotto
• Alexandra Cristea

2
Outline

• Motivation
• ADAPT dimensions
• ADAPT dimensions explained
• Example systems described
• Conclusions
• Information

3
Motivation

• vital aspect for AEH design.
• 1st step towards common design patterns,
• leading to better, semantically enhanced
  authoring systems for AEH.
• easier access to these special tools for
  personalization for the domain expert without
  computer knowledge
• High-level taxonomy proposed
• result of brainstorming sessions EC project ADAPT

4
ADAPT dimensions

• Context of Use
• Content Domain
• Instructional/ Educational Strategy
• Instructional View Detection Mechanism
• Learner Model Adaptation Model
• Presentation Model

5
DESIGN Space
REQUIREMENTS Space
Detection Mechanism
Adaptation Model
context of use instructional strategy .
Feedback
Instructional Views
Learner Model
Content domain
6
Context of use
7
Content domain

• answers to the questions
• what are the proper concept types? (For example
  Fact, Phenomenon, Principle, Example, Formal
  Definition, Informal Definition, Procedure (how
  to do), Process, Hands on, Theory,
  Demonstration, Quotation, Simulation..) and what
  are the proper types of domain relationships that
  are useful for the application purposes? (For
  example explanation for, application of, etc.)

8
Instructional views

• An instructional view is a schema-based
  description of the application, tuned to the
  needs of a user as described by the stable macro
  attributes introduced above. They offer the
  learner a personalized view on the application,
  based on a selection (and restructuring) of the
  content types and relationship types that are
  more appropriate for their motivations,
  high-level goals, background knowledge, learning
  style etc. The instructional view may both filter
  the content design according to some pedagogical
  strategies (defined at requirements level) and
  superimpose some structure on the content domain
  (such as, introducing instruction-oriented
  relationships such as is a prerequisite of)
  which implement a specific instructional
  approach, or meet some specific macro attitudes
  of the user.

9
Detection mechanism

• This design dimension has to do with the
  following aspects Which learner attributes
  (among the ones represented in the user model)
  are detected? When are they detected? (e.g., at
  the beginning of a session, at the end of a (set
  of) session(s), during a session,) What is the
  degree of user/system control on learner model
  state (for which learner attributes)? How are the
  values of user model attributes measured? For
  this latter question, we can envision a number of
  different solutions, such as explicit input of
  some learner attributes (e.g., background
  knowledge, learning preference,), explicit input
  of indirect learner parameters such as the ones
  that can be collected, e.g., via Learning Styles
  Surveys or Questionnaires, assessment tests
  system evaluation of indirect learner
  parameters, based, for example, on the analysis
  navigation behaviour (pages or links used by
  the learner), time spent on given concepts
  (pages or groups of pages).

10
Learner Model

• This design answer to the questions what are the
  relevant learner attributes that should be
  captured by the user model? How do they relate to
  the content? The first question can be further
  decomposed into a number of sub-questions,
  including
• Which attributes are more stable (i.e., have no
  or a low degree of variability during a session
  of use) and which are variable? Which attributes
  have impact on general macro design properties of
  the application and which impact on fine grained
  aspects (micro attributes)? Example of common
  user model attributes are the user knowledge
  about given concepts at different levels this
  can be considered a micro and variable attribute
  for detailed domain concepts levels and a macro
  attribute ( with a lower degree of variability )
  for high levels concepts background knowledge
  outside the application domain a stable macro
  parameter motivations and high-level goals
  stable macro parameters tasks a variable micro
  parameter learning style the user preferences
  on how to perceive and process information a
  stable macro parameter domain independent
  aspects e.g., learning style, personality, sex,
  physical and psychological abilities or
  disabilities micro and stable parameters time
  of use per (set of) concept(s) variable micro
  parameter (in most cases) performance (measured
  though assessment) this has multiple levels of
  granularity and can be consider both a macro and
  a micro parameter, and largely variable amount
  and type of scaffolding required (macro parameter
  at the beginning of a learning experience, may
  become a micro parameter)
• How do attributes relate to the content domain?.
  This can be, e.g., via overlay model (1-N mapping
  from concepts to user micro attributes), or
  independent attributes.

11
Adaptation Model

• Adaptation design concerns the rules which model
  the systems adaptive or adaptable behaviour. The
  design decisions can be organized along two
  dimensions
• Adaptation scope Which hypermedia design
  dimensions are affected by adaptation, among the
  following Content, Navigation, Interaction,
  User Activities and Layout?
• Granularity of adaptation in the large or in the
  small. In the large rules affect the
  instructional views, i.e., determining changes at
  schema level. In the small rules operate within a
  specific instructional view and define fine
  grained changes, of individual instances of nodes
  and links.

12
Feedback Mechanism

• This aspect concerns the design of the dialogue
  established by the system to notify the user of
  changes of the user model, of instructional view,
  or of fined grained aspects such as specific link
  or content structures.

13
ELM-ART ELM Adaptive Remote Tutor
14
TANGOW Task-based Adaptive learner Guidance On
the WWW
15
ISIS-Tutor An Intelligent Learning Environment
for CDS/ISIS Learners
16
Conclusions
In this paper we have described the basis of
taxonomy for the design of adaptive and adaptable
hypermedia, that has been created within the
European Community project ADAPT. The design
dimensions identified here have the role to force
implementers to make their embedded knowledge
about AEH systems explicit, on one hand, and on
the other hand, make it easier for authors to
create their own AEH applications. To test these
dimensions, we are creating an adaptive pattern
language based on them, and using it for the
implementation of some new authoring tools (MOT),
as well as extensions of some old ones (AHA!,
WHURLE). We believe that it is extremely
important to have a common way of expressing the
components for the design and authoring of AEH
for other, related domains, as well, such as
collaborative authoring and open hypermedia. For
the first, authors that have delimited their
tasks precisely can collaborate without major
deadlock situations, being almost independent
from each other. In this way, domain experts can
be working together with pedagogical experts, for
instance, each of them at their own part. For the
second issue, open hypermedia has to rely heavily
on commonly accepted metadata structures and
protocols, in order to be able to reuse data from
outside the space delimited by the current
learning environment. Therefore, this is another
research and application area that can benefit
from such clear identifications of design
dimensions and patterns.
17
Information

• The work is supported by the ADAPT Minerva
  Project 101144-CP-1-2002-NL-MINERVA-MPP.
• Information
• http//wwwis.win.tue.nl/acristea/HTML/Min
  erva/