ESWC2011 Summer School: Front-end to the Semantic Web
1. “interface is the message”
on the path to a usable & personal Semantic Web
Lora Aroyo
VU University Amsterdam
@laroyo
Wednesday, June 1, 2011 1
2. o utline
front-end to semantics: how do we interact with SemWeb Apps?
personalization: what do we need to adapt to users?
example applications: what good & bad is out there?
evaluation: why is continuous evaluation so important?
Wednesday, June 1, 2011 2
3. why interfaces?
invisible computers
multitude of interaction modes
context-sensitive apps
networked devices: bridges between virtual & physical worlds
GUI become central
constantly increasing competition
Wednesday, June 1, 2011 3
4. take ho me message
combine content semantics with user context
integrate seamlessly physical & web worlds
identify relevance to user to rank & select information to present
continuous feedback cycle: to and from user
you need to deal with GUI on configuration level
perform continuous user testing
use real world data
Wednesday, June 1, 2011 4
5. “interface is the message”
Aaron Koblin: Artfully visualizing our humanity, TED Talk, 2011
Wednesday, June 1, 2011 5
6. FRONT-END TO SEMANTICS
how do we interact with the SemWeb Apps?
Wednesday, June 1, 2011 6
8. semantics: what’s special?
explicit semantics (often from open sources, e.g. LOD) used
for system decisions and results
use facetted presentation, searching and browsing of
information
use typically classifications, typologies or other structures of
concepts
integrate data from different sources
aggregate data
Wednesday, June 1, 2011 8
15. PERSONALIZATION
what do we need to adapt to us?
Wednesday, June 1, 2011 15
16. the user matters
when we consider interaction & interfaces, then the user plays a
key role
for good interface design, a good characterization of the user
is needed
first, some concept from theory and literature
Wednesday, June 1, 2011 16
17. user profile
Definition: A ‘user profile’ is a data structure that represents a
characterization of a user (u) at a particular moment of time (t)
So, a user profile represents what (from a given (system)
perspective) there is to know about a user.
The data in a user profile can be explicitly given by the user or
have been derived.
Wednesday, June 1, 2011 17
18. user characteris tics
Personal data
Friend and relations
Experience
System access
Browsing history
Knowledge (learning)
Device data
Location data
Preferences
Wednesday, June 1, 2011 18
19. user mo del
Definition: The ‘user model’ contains the definitions and rules
for the interpretation of observations about the user and about
the translation of that interpretation into the characteristics in a
user profile.
So, a user model is the recipe for obtaining and interpreting user
profiles.
Wednesday, June 1, 2011 19
20. user mo deling
Definition: ‘user modeling’ is the process of creating user
profiles following the definitions and rules of the user model.
This includes the derivation of new user profile characteristics
from observations about the user and the old user profile based
on the user model.
So, user modeling is the process of representing the user.
Wednesday, June 1, 2011 20
21. stereotyping
Stereotyping is one example of user modeling.
A user is considered to be part of a group of similar people, the
stereotype.
Question: What could be stereotypes for conference participants
(when we design the conference website)?
Wednesday, June 1, 2011 21
22. user-adaptive system
Definition: A ‘user-adaptive system’ is a system that adapts itself to a
specific user.
Often, a user-adaptive system (or adaptive system, in short) uses user
profiles to base its adaptation on.
So, designing an adaptive system implies designing the user modeling.
Wednesday, June 1, 2011 22
23. user adaptation
User-adaptation is often used for personalization, i.e. making a
system appear to function in a personalized way.
Question: What user profile characteristics would be useful in
personalizing the conference’s registration site?
Question: How would you obtain those characteristics?
Wednesday, June 1, 2011 23
24. examples: user adaptation
Device-dependence
Accessibility (disabilities)
Location-dependence
Adaptive workflow
Question: Can you give concrete examples for interface adaptation,
both the adaptation effect as the prior user modeling necessary?
Wednesday, June 1, 2011 24
25. adaptive hyperme d ia
Well-studied example of adaptation is ‘adaptive hypermedia’: a
hypertext’s content and navigation are then adapted to the user’s
browsing of the hypertext.
Wednesday, June 1, 2011 25
27. d ialog principles [Grice]
Be cooperative
Be informative
Be truthful
Be relevant
Be perspicuous (be clear)
Wednesday, June 1, 2011 27
28. UI principles [Shnei der mann]
Strive for consistency
Enable frequent users to use shortcuts
Offer informative feedback
Design dialog to yield closure
Offer simple error handling
Permit easy reversal of actions
Support internal locus of control
Reduce short-term memory load
Wednesday, June 1, 2011 28
29. usability heuristics [Nielsen]
Visibility of system status
Match between system and real world
User control and freedom
Consistency and standards
Error prevention
Recognition rather than recall
Flexibility and efficiency of use
Aesthetic and minimalist design
Help users recognize, diagnose and recover from errors
Help and documentation
Wednesday, June 1, 2011 29
30. all abo ut the user’s perspective
modeling the user: what are user’s preferences, interests, history,
activities, etc.
modeling the user’s context: e.g. location, time, device
which of all the data available is relevant
for this user in this context
also called context-aware
Wednesday, June 1, 2011 30
31. user’s context d is tribute d
switching between one context and another
doing things not only for him/herself, e.g. buying present for a
girlfriend
Wednesday, June 1, 2011 31
33. interaction mo des
search, e.g. keyword, faceted
browse, story lines, narratives through collections
annotations of multimedia, e.g. (collaborative) tagging, professional
annotation of text, images and video, tagging games
explanations, hints, user feedback, e.g. explanation of
recommendation results, explanation of autocompletion suggestions
Wednesday, June 1, 2011 33
34. typical examples
recommendation systems, e.g. movies, music, art
user statistics and analysis, e.g. user usage data, profile, group
profiles, etc.
social networking
Wednesday, June 1, 2011 34
35. reco m mender systems
Definition: A ‘recommender system’ is a system that recommends to
a user, based on her individual interests, items that the user could find
interesting.
Examples: music, movies, people, restaurants
Types: collaborative (reason about similar users), content-based
(reason about similar items)
Problems: new users, new items, sparsity, gray sheep
Wednesday, June 1, 2011 35
36. reco m mender systems
movies & TV programs, e.g. Netflix, MovieLens, TiVo, personalized TV
guides
music, e.g. LastFM, Pandora, iTunes Genius
food & tourism, e.g. guides adapted to location, current time, preferences
news, e.g. Google reader, news filters
e-shopping, e.g. Amazon’s recommendations
advertisement, e.g. Facebook personalized ads
art, museums, e.g. personalized search, personalized museum guides
Wednesday, June 1, 2011 36
37. consi derations
Collection of activities/context/attention data
Derive interests from this data
Recommender-specific problems, e.g. cold start, over-specialization
Surface items of interest in the ‘long tail’
Cross-domain recommendations
Multi-person recommending
Granular control for users
Wednesday, June 1, 2011 37
38. user profiles & stats
overview of user preferences, e.g. settings, privacy
overview of user interests, e.g. ranking of interests, links to content
overview of user/group activities, e.g. per topics, per activity, per
date, over a period, overall
comparative views between users, e.g. LastFM, livingSocial movies
user similarity, Twitter similar users to you
different views/visualization over the same set of user data
Wednesday, June 1, 2011 38
41. social networking
professional networks & events, e.g. LinkedIn, Mendeley
people, organizations, e.g. Facebook, MySpace
Twitter
social bookmarking, e.g. Delicious, StumbleUpon, Diggit
GetGlue
Books, e.g. LibabryThing
Wednesday, June 1, 2011 41
42. EXAMPLE APPLICATIONS
Interfaces & Personalization on SemWeb
Wednesday, June 1, 2011 42
58. personalized experience
Personalized
Web
Access Online
Tour
Wizard Personalized
Mobile
Tour
Interactive tours
Semantic Search
Interactive user modeling
On-the-fly adaptation
Museum tour maps
Recommendations of
artworks & art topics Synchronized user
Historic timeline
profile
Wednesday, June 1, 2011 58
69. dynamic adaptation
For each artwork in the museum:
Related works
Include in the tour ( & recalculate the map/tour)
Indicate relevance in terms of e.g. personal interest, position, recommended by friends, by Rijks, on view
Rate to indicate interest
At any point of the tour:
Include/exclude artworks
Adjust tour length
Change navigation in and outside of the tour
Save for other tours
Wednesday, June 1, 2011 66
70. EXAMPLE 2
professionals vs. lay users on Web 2.0
semantic annotation of Rijksmuseum prints
http://e-culture.multimedian.nl/pk/annotate?
semantic tagging: http://waisda.nl
Wednesday, June 1, 2011 67
71. Autocompletion with multiple
vocabularies
http://slashfacet.semanticweb.org/wordnet/search
http://slashfacet.semanticweb.org/autocomplete/demos/
Wednesday, June 1, 2011 68
87. watching TV in a group
Environment Age
Interact with the second 15 - 35 years old
screen as a group
Friend interaction at home Type of Activities
Watching as a group quiz and betting games
change camera view
Synchronization information regarding the
TV & Second Screen content of the program
between second screens textual captions
between second screens &
TV show content provider Type of Program
Sports
Wednesday, June 1, 2011 81
88. observations
for more details check out our blog at http://notube.tv
Wednesday, June 1, 2011 82
89. observations
for more details check out our blog at http://notube.tv
Wednesday, June 1, 2011 83
90. second screen & TV
functionalities
shared virtual space synchronization with second
voice dubbing screen
subtitles “overlay” on top of the main
related information TV-picture
quizzes censoring
voting & betting different camera views
scene-grab & share group alerts
social interaction
live-chat
parental advisory
uncensored version
different camera views
Wednesday, June 1, 2011 84
92. CHIP users
Target users’ characteristics
small groups with 2-4 persons and a male taking the leading role
(67%)
middle-aged people in 30-60 years old (75%)
higher-educated (62%)
no prior knowledge about the Rijksmuseum collection (62%)
visit the museum for education (98%)
Wednesday, June 1, 2011 86
94. contextual analysis
Context
ual obse
rvations
Define familiarity with the
domain
s Define familiarity with
iew collections/vocabularies
ter v
r in
Use
Va Identify use cases
lid
ate
Identify navigation patterns
sks
Model user’s ta Identify requirements for
user groups
Wednesday, June 1, 2011 88
105. take ho me message
combine content semantics with user context
integrate seamlessly physical & web worlds
identify relevance to user to rank & select information to present
continuous feedback cycle: to and from user
you need to deal with GUI on configuration level
perform continuous user testing
use real world data
Wednesday, June 1, 2011 97