The Language Manager

1
(No Transcript)
2
Testing Web Navigation for all an Agent-based
Approach

• Presented by the

The Human Communication and Interaction
Research Group University of Oviedo, Asturias,
Spain www.hci.uniovi.es
3
A Simple Question?

• Which is the best location for the navigation bar
  of a web site?

News Events Docs Surveys
News Events Docs Surveys
4
Navigability Testing

• There is not a clear way to predict how the users
  navigate using a specific hypermedia device.
• There are too several cognitive factors involved
  not under the control of the researcher.
• How can designers decide which Navigation Model
  better adapts to the user mental model?
• Prototyping Usability Testing

5
Some Disadvantages

• Requires permanent observation.
• Each volunteer needs at least one observer.
• It is an expensive and slow process.
• Size of the samples are generally small due to
  economical reasons.
• Low reliability of the testing results.
• It is quite difficult to discover a new kind of
  not already identified users.
• Volunteers know that they are being observed
• Knowledge about the experiment influences
  navigability.
• Destruction of spontaneous behaviour (user own
  navigational metaphor).

6
Remote Testing

• Volunteers do not know the role they are playing
  during the experimental process.
• User feels free to explore the Navigational
  Models.
• Designers are not be present while the users
  explore the navigation model.
• Testing is not affected by external factors.
• Usability testing performed in the same
  environment (conditions) under which the final
  product will be used.
• Simplifies the process of detecting new
  navigational metaphors.
• Information about how the user-computing
  environment affects navigation can also be
  observed.

7
Goals of ANTS

• ANTS Automatic Navigability Testing System
• Supports Navigability Testing under the ideal
  situation described.
• Observe users as they navigate through hypermedia
  artefacts.
• Silent recording users do not notice how they
  are being observed.
• Automatic detection of special interaction
  requirements.
• User navigates using only the keyboard?
• What is the users accuracy when using a pointing
  device?
• Do exist collection of objects that are never
  used by specific users?

8
The Anthill Metaphor

• Based on the client-server paradigm.
• Design metaphor A community of ants.
• ants agents.
• anthill server.
• food information, user behaviour.
• picnic user navigation session.
• anthill warehouse databases or files.

9
The Anthill Metaphor
User
SERVER
10
Information Collected

User Model
ANTS
11
Information Collected

• Navigation Profile
• Complete identification of the node visited by
  the user.
• Time of arrival.
• Time of departure (time-delay navigation
  strategies).
• Device Profile
• Internet connection speed
• Device Performance.
• User Profile
• Set of components selected by the user.
• Window locations clicked on.
• Visible and hidden components during scroll
  tasks.
• User typing skills.
• User mouse motion skills.

12
A Simple Question?

• Which is the best location for the navigation bar
  of a web site?
• What is the designers preferred location?

13
A Simple Experiment

• Single node with TWO navigation bars (left and
  hand side located).
• Colours, shapes and contents are the same in both
  bars.
• Every bar has only a link.
• Software agents observe users first selection as
  well as his/her reaction time (RT).

LINK
LINK
14
Results
Universe 342 navigation sesions. Reaction Time
Average (RT) 23.94 seconds.
Left 179 times (52.5) RT 23.83 seconds.
Right 163 times (47.5) RT 24.08 seconds.
15
Remarks

• Static design addressed the interaction
  requirements of only the 50 of the users.
• Left handed users preferred the left hand link.
• Right handed users preferred the right hand link.
• Processing of a printed publication requires
  Perception only.
• Interactive media requires both, Perception and
  the Human Motor Processor.

16
Conclusions

• When using a Remote Testing Approach
• There is no need to assign human resources to
  conduct user observation.
• More Navigation Sessions can be observed using
  limited human resources.
• Testing quality improves as the size of the
  samples can be increased with no cost.
• Navigation takes place in the user-computing
  environment.
• There is no need to assign laboratory resources
  to Navigability Testing.
• Navigation behaviour is not altered by external
  factors.

17
Thanks!
Human Communication and Interaction Research
Group University of Oviedo Departamento de
Informática C/Calvo Sotelo s/n 33007 OVIEDO,
Asturies Spain www.hci.uniovi.es ANTS/GADEA
won the 2005 Edition of the TAW Awards as the
best Research Project on Web Accesibility.
ANTS/GADEA is developed through Project
IB05-152 SEELE Automatic generation of Web
Interfaces dynamically tailored to the
requirements of different kinds of disabilities,
funded by the Asturian Office of Science and
Technology (FICYT).
CALL FOR PAPERS e-Minds International Journal on
Human-Computer Interaction Issue 2,
2006 Submission Deadline Sept 12, 2006.
18
Big Brother

• Notice that
• Volunteers know they are participating in an
  Usability Testing.
• The only fact they ignore is the observation
  technique used.
• The technique used will be revealed to users once
  the experiment will have concluded.