Upload
jaceksg
View
214
Download
0
Embed Size (px)
Citation preview
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
1/30
Personalization for Individual DifferencesPersonalization of Interactive Information Systems
to Match Cognitive Differences Among People
http://www.scils.rutgers.edu/~jacekg/
http://www.gwizdka.com
Department of Library & Information ScienceSCILS
Rutgers University
New Brunswick, NJ, USA
PhD 601 - Dec 3, 2008
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
2/30
Background
Electrical engineering
Software development
Information systems
Human factors / industrial engineering
Human-computer interaction
Jacek Gwizdka: http://www.scils.rutgers.edu/~jacekg/
http://www.gwizdka.com
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
3/30
Research Interests
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
4/30
personalization is multi-dimensional
Facetsof personalization
Relevance/interest
Task
Problem state
Personal characteristicscognitive characteristics, affect,
education, demography
Personal preferencesinteraction styles, info organization
Context/situationlocation, urgency
(Belkin, 2006)
What can be personalized
content
source selection; genre
presentation
results; documents
interaction
search UI; navigation
help
form & time of delivery
user interaction: effective, efficient, pleasurable
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
5/30
Overview
Cognitive differences
relevance to information interaction
implications on the design of information systems
approaches that take them into account
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
6/30
Individual Differences - Example 1
Cognitive differences and
reading
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
7/30
Example 1
Moud a text-ouly sight bee ideale for soweoue mith a reabing
bisorber? Harblee, Iwages are uot dab for accessaabilledea. They
actnally iucreese cowqreheusiou aub nsadililte for wost anbleuces.
Mhat wuay qeiqle bo uot kuom, throngh, it there is wuch mor at
the accessability for au iwage theu jnst its alt text. Sowe qeople
mroughly assnwe that iwages are dab for accessedilite, siuce alt
text esseutially reqlaces the iwage mith a text-ouly versiou of thatiwage.
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
8/30
Example 1
Would a text-only site be ideal for someone with a reading disorder?
Hardly, images are not bad for accessibility. They actually increase
comprehension and usability for most audiences.
What many people do not know, though, is there is much more to
the accessibility of an image then just its alt text. Some people
wrongly assume that images are bad for accessibility, since alt text
essentially replaces the image with a text-only version of thatimage.
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
9/30
Individual Difference Example 1
Dyslexia
difference in perceptual and cognitive processing
Moud a text-ouly sight bee ideale for soweoue mith a
reabing bisorber? Harblee, Iwages are uot dab for
accessaabilledea. They actnally iucreese cowqreheusiou
aub nsadililte for wost anbleuces.
Mhat wuay qeiqle bo uot kuom, throngh, it there iswuch mor at the accessability for au iwage theu jnst its
alt text. Sowe qeople mroughly assnwe that iwages are
dab for accessedilite, siuce alt text esseutially reqlaces
the iwage mith a text-ouly versiou of that iwage.
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
10/30
Individual Differences - Example 2
Cognitive differences and
information search
in web directories
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
11/30
Individual Difference Example 2
Cognitive Style: FD / FI: field-dependence / independence
Witkin et al. (1971)
FD FI
holistic perception (whole objects) analytic perception (parts)
global focus focus on detail
external references internal references
passive in locating information active in locating information
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
12/30
Individual Difference Example 2
Cognitive Style: FD / FI: implications for information systems
Witkin et al. (1971)
FD FI
externally imposed structure own structure
extra guidance locate info directly
subject organization
breadth (more main cats, less sub-cats) depth (less main cats, more sub-cats)
separate category levels categories together
sorted by relevance alphabetical organization
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
13/30
Example 2
FD
FI
From: Chen, S. Y., Magoulas, G. D., & Macredie, R. D. (2004). Cognitive styles and users responses to structured
information representation. International Journal on Digital Libraries, V4(2), 93-107.
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
14/30
Example 2 Subject Categories
FD
FI
From: Chen, S. Y., Magoulas, G. D., & Macredie, R. D. (2004). Cognitive styles and users responses to structured information
representation. International Journal on Digital Libraries, V4(2), 93-107.
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
15/30
Individual Differences - Example 3
Cognitive differences and
information scanning
(in email)
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
16/30
Individual Differences In Email Use
Email Avoidance Email Addiction
You should check your e-mail moreoften. I fired you over 3 weeks ago.
The 12-step program to over-coming addiction to e-mail.
I will not:- check my email
more than twice a day- skip meals to spend
time sending email
- base my self-worth onthe no of emails I receive
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
17/30
Example 3 - Scanning Email Messages
Scanning Task: find message in inbox based on partial header info
Differences in cognitive abilities: working memory WM, visual memory
VM, flexibility of closure CF
UI-Visual UI-Text
(Gwizdka, CASCON2002, PhD2004, Interacting with Computers2004)
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
18/30
Example 3 - Scanning Email Messages
better visual memory(mv1 & mv2)
less scrolling
better working memory(wm)
less sorting
betterflexibility of closure
(cf2)
more scrolling
CF
WM
MV2MV1
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
19/30
Individual Differences - Example 4
Cognitive differences and
information keeping
(in / around email)
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
20/30
Example 4 the Keepers and the Cleaners
Is there a difference in in email habits?
Esp. in how people handle todo/task-related messages
Cognitive difference - flexibility of closure (CF)
FD ~ low CF FI ~ high CF
holistic perception (whole objects) analytic perception (parts)
global focus focus on detail
external references internal references
passive in locating information active in locating information
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
21/30
Example 4 the Keepers and the Cleaners
1- The Cleaners: transfer todos from email
2- The Keepers: keep todos in email
Email Habit Variablesdetermining clusters
The CleanersCluster #1
The KeepersCluster #2
When email is read at specific times all the time
Email interrupts other tasks no yes
Uses search in email no yesKeeps events in email no yes
Keeps to-do's in email no yes
Emails self-reminders yes no
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
22/30
Example 4 the Keepers and the Cleaners
Flexibility of closure (CF) differed between the Keepersand the Cleaners:
Low CF The Cleaners
High CF The Keepers
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
23/30
Individual Differences - Example 5
Cognitive differences and
information search
(difference search engines)
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
24/30
Jacek Gwizdka 24
Individual Differences Example 5
plainresult list
faceted search - ALVIS
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
25/30
Results: Person (ID) & interactions
Working Memory (WM) task performance on ALVIS, but noton Googlehi-WM more search effort on ALVIS (more pages, more
bookmarks, spent more time) than on Google
lo-WM less effort on ALVIS than on Google
Google ALVIS
searcheffort
high WM
low WM
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
26/30
Yes, Individual Differences - So What?
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
27/30
Yes, Individual Differences - So What?
Research questions
1. relationships among: user characteristics,
info presentation and user performance, preference,
perception effective, efficient, pleasurable
2. identification of user types
3. identification of user states
e.g., overloaded, lost
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
28/30
Yes, Individual Differences - So What?
Approaches:
accommodate different users in one interface
provide alternative interfaces for different users
create interfaces that adapt to users
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
29/30
Identify User: Rich Data Collection
User model.Interaction pattern processing, etc. (e.g.machine learning)
Observable useractions
(including data from input
devices, e.g., mouse pressure)
Optionalintervention
System
Physiologicaldata (e.g. EEG)
Notification
Eye-tracking data
Identified Usertype or state
8/3/2019 JGwizdka_PhD601_2008-12-03-sh
30/30
Jacek Gwizdka 30
Thank You
Questions?Jacek Gwizdka
Dept. of Library & Information Science
Rutgers UniversityNew Brunswick, NJ, USA
http://www.scils.rutgers.edu/~jacekg/
http://www.gwizdka.com
This research was partially funded by a grant from IMLS: LG-06-07-0105-07
Personalization of the Digital Library Experience