100 likes | 401 Views
Definition:. Intelligent user interfaces are human-machine interfaces that aim to improve the efficiency, effectiveness, and naturalness of human-machine interaction by representing, reasoning, and acting on models of the user, domain, task, discourse, and media.. IUI range of approaches:. Personali
E N D
1. Intelligent User InterfacesPattie MaesFluid Interfaces Group
2. Definition: Intelligent user interfaces are human-machine interfaces that aim to improve the efficiency, effectiveness, and naturalness of human-machine interaction by representing, reasoning, and acting on models of the user, domain, task, discourse, and media.
3. IUI range of approaches: Personalized, proactive interfaces
Personalization
User modeling/user profiling
Adaptive interfaces (machine learning)
Recommender systems
Software agents
…
Interfaces to mimic human-to-human interaction
Conversational interfaces, natural language techniques in interfaces
Multi-modal input and output
Personified, anthropomorphic interfaces
Interfaces that recognize affect
Interfaces with common sense
Interface that plan and reason
Main conference: annual IUI conference
4. Concept Videos Knowledge Navigator, Apple, 1987
http://www.youtube.com/watch?v=3WdS4TscWH8
Future Shock, Apple, 1988
http://www.youtube.com/watch?v=QcYrPkFe2J0
Starfire Video, Sun, 1994
http://asktog.com/starfire/starfire.mp4
…
5. Contentious Issue #1 Is personification good or bad?
Microsoft Bob: http://www.youtube.com/watch?v=ZegWedG-jk4
Microsoft Clippy: http://en.wikipedia.org/wiki/Office_Assistant
Microsoft’s Ms Dewey: http://www.youtube.com/watch?v=QPDvfPb2SqE&feature=related
Nicole: http://www.youtube.com/watch?v=Sgg-WkfSOxY
Multi-touch Barcelona “real human interface”: http://vimeo.com/4697849
6. Contentious Issue #2 Is the problem “AI complete”?
7. Contentious Issue #3 Trust & predictability?
awareness of functionality
predictability of outcome
understanding limitations
user should be in control
reporting/explanation by the system
able to turn off, modify system
8. Contentious Issue #4 Privacy?
user-ownership of data/agents
user determines what is made available and to whom
9. Contentious Issue #5: Cost/benefit tradeoff
High likelyhood of being relevant/useful to the user
offered as unobtrusive as possible
Minimize disruption, distraction
Ignorable (ramping interface)
requires minimal user effort to access
Challenges in:
Natural, “on-the-move” interfaces
10. Techniques Used in Profile Building & Recommendations Explicit user input of preferences, interests
Statistical & data-mining approaches: recognizing patterns in data of:
one user
many users
Using background knowledge (common sense knowledge) to infer/learn/generalize
11. Intelligence in commercial applications Recommendation systems are ubiquitous
Typo & grammar correction
Automated bidding
Spam filters
More, better speech interfaces
…