Roth, Emilie M., Malin, Jane T. and Schreckenghost, Debra, L.:
“Paradigms for Intelligent
Interface Design”,
in Helander & al (editors):
“Handbook of Human-computer interaction”, second edition, Amsterdan
1997
Jarmo Parkkinen
Contents
• Definitions for “Intelligent Interfaces”
• Applications and examples
– Greek Oracle
– Cognitive tool and representational aid
– Cooperative systems
• Summary (and future work)
Definitions (1/2)
• Term “Intelligent Interface” refers to
– design of user interfaces for intelligent systems
– design of user interfaces that utilize knowledge
based approaches
• The example categories examined in the
paper handle intelligent interfaces as
– as cognitive tools
– as members of co-operative man-machine
systems
– as representational aids
Definitions (2/2)
• The examples are given “... rather
complementary metaphors that provide
converging insights”
• The definition of “Intelligent Interface
design” in the article is paradigmatic one; it
is not to be understood per se, but an
umbrella term covering wide range of
designs
Greek Oracle (1/2)
• Autonomous
problem-solving
agents
• User
– gathers data
– filters solutions
Problem domain
Data
Gatherer
User
Solution
Filterer
Machine Problem-Solver
G(r)eek Oracle (2/2)
• Problems
– Lack of user acceptance due to user’s filteringrole
– Brittleness caused by unanticipated variability
– Deskilling caused by the roles; human just
filters solutions
– Making the human decision poorer due to the
belief of “superhuman aid”
• Emergent need of co-operative systems and
cognitive tools
Cognitive tool (1/2)
• Machine part
creates new
integrated sources
of information
• Data synthesis
for overcome
data overflow
Problem domain
Multiple information sources
Intelligent Interface:cognitive tool
Information Integrator
User
Cognitive tool (2/2)
• Succesful examples
– better visualization of monitored process
– better information handling capabilities
– conceptualization aids that enable user to search for and
discover patterns in the data
– data (disease) profile summaries (in fast and efficient
manner)
• Problems
– when integration is not good, they provide yet another
information channel
Representational aids (1/2)
• External
representations
• Capitalize
advantages in
graphics and
knowledge-based
techniques in
computer-generated
presentations
Problem domain
Intelligent Interface: representational tool
Automatic Presentation Design
User
Representational aids (2/2)
• External presentations aid
–
–
–
–
Problems structuring
Mentally economic forms of processing
By reducing memory load
By directioning attention
• Current research focuses on efficient
– identification of relevant data characterics to create
presentation and communicative purpose
– mapping between data and graphics; information type
and media
– coordination of media, ie. graphics and text
Cooperative Systems (1/5)
• Based on
metaphor of
cooperative
teams
• Generate
Problem domain
Shared external representation
– common
ground
– shared frame
Intelligen
of reference Machine Agent
Cooperative
Tasks
User
Cooperative Systems (2/5)
• Critiquing systems
• Team players as
–
–
–
–
followers
reliable participator
coordinator
communicator
Cooperative Systems (3/5)
• Passive and active critics
– passive: spelling and grammar check
– active: hints and automatic corrections of splelling
errors
• Systems that have and/or allow user build a
database of guiding information. Information is
given as
–
–
–
–
hints of solution and critics to user solutions
influencers: guidance before or during the task
debiasers: alert and suggest a correction if user errs
directors: wizard
Cooperative Systems (4/5)
• Team player systems
– reduce workload of human participant
– automates time-critical tasks
– Provide ”catalyzing” information to ”nudge” difficult
process forward
– Takes handover (from or to human) well
– Enables human team members to have shared
knowledge of eahc others actions and situation
– Is effective communicator, giving only the relevant
information and requiring only the most relevant input
– Provides interruptions to share context and prevent
errors
– Sees when information was / was not understood
Cooperative Systems (5/5)
• Requirements and critics
– Human members must have ability to shut automated
system down and to see which parts are shut down
– Intelligent software must ”defer” to human
– Human must be able to understand the system – many
systems perceive human disgreement as failure in
understanding, when common ground should be sought
– Retrospective dialogues give transparency to the
underlying system, but do not suit in time-critical
environments
– Responsibility is still left to human
Summary
• ”Intelligent interface” is a design paradigm of user
interfaces for intelligent system and knowledge
based solutions
• The intelligent interface works as
– cognitive tools
– representational aids
– memeber of cooperative person-machine systems
• User must be seen as the central actor
• Intelligent interface supports user’s cognitive
process