1. User-Adaptive Systems
CSA3212:
User-Adaptive Systems
Topic 6: Adaptation Techniques
Dr. Christopher Staff
Department of Intelligent Computer Systems
University of Malta 1

2. Aims and Objectives
We will look at stages in the adaptation process and revisit the problems that AHS/UAS try to solve 2

3. Stages in the Adaptation Process
Collect data about the user
Process data to build/update user model
Apply user model to provide the adaptation
Last stage always fully automated in adaptive hypertext systems
1st and 2nd stages may require user intervention. Ideally, fully automated 3

4. Stages in the adaptation process
If user and system need to co-operate to achieve adaptivity, then it is called collaborative user modelling
From [Brusilovsky96] 4

5. What can be adapted?
A hypertext is a collection of nodes that are connected by links
So what can be adapted?
The node content and user interface - adaptive presentation
The node organisation and hyperspace representations (e.g., index, map, overview) - adaptive navigation 5

6. Generic approaches
Adaptive Presentation
Adaptive Navigation 6

7. Adaptive Presentation
Anything to do with adapting how the presented material is displayed to users
Examples
Adapting the UI (more to do with adaptive user interfaces than adaptive hypertext per se)
Modify data presentation so that it is presented in user's preferred choice (e.g., charts) 7

8. Adaptive Presentation
More examples
Automatically expand/collapse glossary items according to user's level of expertise
Provide "trails" through hyperspace according to user's level of expertise
Can be fairly complex - if user does not know concept A, then unlikely to know concept B, so concept B should also be automatically expanded 8

9. Adaptive Presentation
Adaptive presentation can be useful in any adaptive hypertext system, but mostly used in Intelligent Tutoring Systems
e.g., MetaDoc 9

10. Adaptive Navigation
Focused around implicit link types in Web- based systems (ideally, link types are explicit)
E.g., Letizia, WebWatcher, Personal WebWatcher, HyperContext
Already identified implicit link types in previous lecture…
navigational, context-free, context sensitive, index, TOC, glossary, map, ... 10

11. Problems Revisited Lost in HyperSpace Cognitive overload
Complexity of the search space
Search-browsing
Static hypertext structure 11

12. Lost in HyperSpace
Mark-up nodes to include link to landmark which user knows about
Show relationship between landmark node and current node
Reduce cognitive load on user 12

13. Cognitive overload
Move (some) linking functions into AHS system by, e.g., providing "See Also" links
Automatic/dynamic linking
Reduce number of outlinks by removing links to non-relevant nodes
Link recommendation services
Add explanations/remove extraneous text 13

14. Complexity of the search space
Removing links reduces the size of the search space
Recommending links assists the user in deciding which links are likely to lead to relevant information
Automatically modifying content enables readers to understand concepts at their level of understanding 14

15. Complexity of the search space
Learn from user access paths to re-organise hyperspace
Learn to associate user terminology with document content 15

16. Search-browsing
Dynamic linking as a result of where user is combined with any stated query terms
Recommend links (paths) to relevant documents 16

17. Static hypertext structure
Users can modify structure, and/or system can learn from users
What change(s) to hypertext systems would be required? IR systems?
Dynamic hypertext systems 17

18. Adaptation Technologies18

19. Direct Guidance Mainly through "Next" buttons!
Can include link/path recommendation
What are the pre-requisites? 19

20. Adaptive Sorting of Links
Ordering links according to some ranking scheme
On what basis?
On which of the implicit link types can this be provided? 20

21. Adaptive Hiding of Links
Hide links which would lead to non- relevant information
What are the pre-requisites? In ITS systems? In generic AHSs?
Identify implicit link types which support adaptive hiding 21

22. Adaptive Annotation of Links
Mark-up anchor text/link description to explain the information at the destination of the link
Description must be adaptive! 22

23. Adaptive Link Generation
Must links be predetermined?
Can automatically discover new links and add them permanently
Must the destination of links be fixed?
Add similarity-based links or recommended links (temporary)
E.g., adapt navigational links to point to landmarks known to user 23

24. Insert/remove Text Fragments
Automatically add or remove text fragments based on predetermined information about user’s knowledge of topics 24

25. Stretchtext
Expand or contract text based on system’s knowledge of what user already knows
Similar to insert/remove text, but different technology
User can also interactively expand/contract text to give feedback to system 25

26. Dimming Fragments
Rather than modifying the presentation of a familiar page (and confusing the user’s mental model of the page), dim rather than hide non-relevant information 26

27. Adaptation of Modality
How should information be presented to user?
Speech, music, animation, video, text?
Assumes that information exist in different forms to begin with! 27