Docstoc

Slides - Department of Electrical Engineering and Computer Science

Document Sample
Slides - Department of Electrical Engineering and Computer Science Powered By Docstoc
					Intelligent User Interfaces and User
         IUI Overview
         User Modeling via Stereotypes

         Models of User’s Emotions
Intelligent User
Interfaces: An
Introduction
                Two Major Areas of
                Discussion


   Introduction to Intelligent User Interfaces (IUI)

   Overview of the collection of readings
    contained in “Readings in Intelligent User
    Interfaces”
                Motivation

   WIMP (windows, icons, menus, and pointing)
    now standard for most apps

   IUIs will provide adaptivity, context sensitivity,
    and task assistance

   IUIs should be learnable, usable, and
    transparent
            Definition

Intelligent User Interfaces are human-machine
  interfaces that aim to improve the effiency,
   effectiveness, and naturalness of human-
     machine interaction by representing,
 reasoning, and acting on models of the user,
      domain, task, discourse, and media.
Definitions

   Mode or Modality refers primarily to the human
    senses: vision, audition, olfaction, touch, and taste.

   Medium refers to the material object used for
    presenting or saving information (I/O devices)

   Code refers to a system of symbols (natural
    language, pictorial language, gestural language)
Relationship Between
Medium, Mode, and
Code
What’s Not Mentioned?

 Human mediums
eyes, ears, skin, etc.
IUI Architecture
Current Interface Pratice
Related to IUI
             No Generation w/o
             Representation

Various constiuents of multimodal communication
should be generated on the fly from a common
representation of what is to be conveyed w/o using
any preplanned text or images.
AI’s Application to UIs
What’s Not Mentioned?

    Affect
   of course
                     IUI’s Roots
   Turing Test
   Pattern matching from a conversational database
   Intelligent Tutoring
   Automated Interface Design
   Natural Language Interfaces
   Standard Reference Model (SRM) for Intelligent Multimodal
    Presentation Systems (IMMPS)
   Email Filters
   Bayesian-based user models
                  Major Topics Covered
                  by Readings

   Analysis of Input
   Generation of Output
   User and Discourse Models
   Model-Based Interfaces
   Agent-Based Interaction
   Empirical Evaluation
                  Resources

http://www.mitre.org/resources/centers/it/maybury/iui99/sld001.htm
Intelligent User Interfaces: An Introduction


http://cslu.cse.ogi.edu/HLTsurvey/
Survey of the State of the Art in Human Language Technology (1996)


http://search.nap.edu/readingroom/books/screen/
More Than Screen Deep: Toward Every-Citizen Interfaces to the Nation's
Information Infrastructure
                  Resources
http://degraaff.org/hci/
HCI Index (news and a list of references)


http://www.etaij.org/
Electronic Transactions on AI


http://www.dfki.de/
The German Research Center for Artificial Intelligence


http://www.dfki.de/sigmedia/
Multi-Language Processing


http://www.i3net.org/
European Network for Intelligent Information Interfaces
User Modeling via
Stereotypes
    Elaine Rich
    University of Texas
               Why User Modeling?
   People need to form a model of the person with
    whom they are dealing before they can behave
    properly.

    Examples:
    PS2 Controller size, Auto Insurance Quote
User Modeling via
Stereotypes
A stereotype is a cluster of characteristics

e.g.
Profession Stereotype:
income level, dress code at work, degree, etc
Ethnic group stereotype:
Food preference, spacial distance
Stereotypes Are Based
on Domain
Entertainment Industry
care:movie, music preference, education level
don’t: dress, weight, height
Car Sales Industry
care: income, family size
don’t: education level, religion
              Stereotypes Are Based on
              Probability
How confident are we ?

Computer Science Professor - NON_TV 85%
College Graduates - over 20 90%



Probability may result in unfairness.
So, a Stereotype is a set of
          triples
(Attribute, Value, Rating)
Medical Doctor
        (Income, 4, 900)
        (Education, 5, 900)
        (Afraid-of-Blood, -5, 900)
        (Watch-TV, -3, 800)
A Stereotype Contains
Triggers
A trigger is a hint that instantiate other stereotypes.
Medical Doctor
       High_Education_Trigger -> NonTV_StereoType
       High_Income_Trigger -> House_Owner


A trigger also has rating as confidence level.
Over View of Grundy the
Librarian
   Has pre-built stereotypes
   Hierarchical memory
    Global, individual, dialogue
   Algorithm for activating stereotypes
   Adaptation of stereotypes
                  Activating Stereotypes (by
                           triggers)
   If trigger already instantiated, do nothing, if
    not, instantiate it.

    Example:
    Name “John” - Man_trigger (instantiate)
    “Father” - Man_trigger (ignore)
                   Activating Stereotypes (by
                            triggers)
   If stereotype has not been activated before, it
    is activated now.
    Name “John” -> Man_trigger -> Man_stereotype
   If stereotype has been activated before and
    still active (confirmation)
    “Father” -> Man_trigger_Rating up ->
    Man_Stereotype_Rating up
                  Activating Stereotypes (by
                           triggers)
   If stereotype has been activated before, but
    it’s not activated now. The situation must be
    re-examined on the basis of the balance of
    the evidence is in favor or opposed to the
    stereotype.

    How to calculate the balance of evidence???
    What’s the definition of “in favor” or “oppose”
    ?
                  Adaptation of stereotypes
   Why?
    Lack of real data at construction time results
    in the need of adaptation of stereotypes.

   Concern
    Prevent bias because of frequent usage by a
    few users, need weighted constant of the
    overall values.
                       Adaptation of stereotypes
   Confirmation Case
    increase VALUE
    increase RATING

    M.D. -> (Income, 3, 800)
    Drive Ferrari -> (Income, 4, 900)

    New_Value = (3 * 100 + 4) / (100 + 1) = 3.01 > 3
    New_Rating = 800 + 900/800 = 801.125 > 800
                       Adaptation of Stereotypes
   Conflict case
    decrease VALUE
    decrease RATING

    M.D. -> (Non_TV_Person, 4, 800)
    Subscribe to TV_Guide -> (Non_TV_Person, -5, 900)
    New_Value = (4 * 100 - 5) / (100 + 1) = 3.91 < 4
    New_Rating = 800 - 900/800 = 799.875 < 800
             Results of Grundy

Controlled   Random

GOOD            102       54


BAD             39         60

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:3
posted:7/15/2011
language:English
pages:32