LocationAware Systems in HumanComputer Interaction

1
Location-Aware Systems in Human-Computer
Interaction

• Loren Terveen Pam Ludford

2
A little about me

• PhD from University of Texas grad student at MCC
  
• 11 years at ATT Labs / Bell Labs
• Came to UMN in April 2002, joined GroupLens
  Research
• Faculty Joe Konstan, John Riedl, and I
• About 15 students, split pretty evenly between
  PhD, MS, and BS
• Historically, core research area is recommender
  systems
• My area Human-Computer Interaction /
  Computer-Supported Cooperative Work
• Specific research interests
• Online communities
• Recommender systems
• Ubiquitous computing ? Ill talk about this more
  today

3
Outline

• Very brief intro to HCI and CSCW
• A look at some location-based HCI work
• My research in the area

4
Human-Computer Interaction

• According to ACM SIGCHI
• Human-computer interaction is a discipline
  concerned with the design, evaluation and
  implementation of interactive computing systems
  for human use and with the study of major
  phenomena surrounding them.
• Interdisciplinary
• Computer Science Psychology Sociology
  Anthropology Graphical Design Industrial
  Design

5
Foundations of UI Design (1)

• Human psychology
• Short-term long-term memory
• Problem-solving
• Attention
• Design principles
• Conceptual models knowledge in the world
  visibility feedback mappings constraints
  affordances

6
Foundations of UI Design (2)

• Understanding users and tasks
• Tasks, task analysis, scenarios
• User-centered design
• Low, medium, and high-fidelity prototypes
• Evaluating designs
• Without users cognitive walkthroughs heuristic
  evaluation action analysis
• With users qualitative and quantitative methods

7
Current HCI topics

• Computer-supported cooperative work
• Online communities
• Ubiquitous computing
• Interfaces for handheld wearable devices
• Interaction Techniques
• Multimodal interfaces / recognition technologies
• Information visualization
• Tangible interfaces
• Universal usability
• Theory
• Design methods
• Social issues trust, privacy
• .

8
HCI at UMN/CS

• 5115 design and evaluation methods
• 5116 implementation, software toolkits
• 8115 graduate-level reading and research
• Collaborative Computing Special Topics class
  this semester I plan to make it a regular class
  in future

9
Computer-Supported Cooperative Work

• HCI when more than one user is involvedit turns
  out lots of work (play, etc.) is collaborative
• Groupware the technology that people use to work
  together
• systems that support groups of people engaged in
  a common task (or goal) and that provide an
  interface to a shared environment.
• CSCW studies the context and use of groupware
• CSCW is the study of the tools and techniques of
  groupware as well as their psychological, social,
  and organizational effects.

10
Email
11
Instant Messaging
12
Shared Web Browsing
13
Shared (Asynchronous) Editing
14
LiveJournal friends page
LiveJournal most recent posts by this users
friends
15
Slashdot
16
Place/time categorization
Meeting rooms, class rooms
In/out boards, Team rooms
Chat, IM, MUDs, text messaging, video
conferencing, NetMeeting, games
Email, online communities, shared documents
17
Relevant HCI work

• Visualization (will cover briefly)
• Focus things you can do once location (of people
  and objects) is known
• Location-based awareness of people
• Association of virtual information with
  real-world places

18
Interactive Map Visualizations

• In HCI work, the key contributions are in the
  area of interaction techniques, not
  visualization/display
• Notable work from the University of Maryland
  starting in the early 90s

19
Dynamic queries / Starfield displays
20
More recent interactive maps
21
(No Transcript)
22
(No Transcript)
23
Location-aware HCI

• It was a map showing every detail of the
  Hogwarts castle and grounds. But the truly
  remarkable thing were the tiny ink dots moving
  around it, each labeled with a name in miniscule
  writing. Astounded, Harry bent over it. A
  labeled dot in the top left corner showed that
  Professor Dumbledore was pacing his study the
  caretakers cat, Mrs. Norris, was prowling the
  second floor and Peeves the poltergeist was
  currently bounding around the trophy room.
  (Harry Potter and the Prisoner of Azkaban, p.
  192-193. J.K. Rowling, 1999.)

24
P3 Systems Jones Grandhi, 2004

• Systems that link (some combination of) People to
  People to Places

25
Their categorization
Where have my buddies been? (social navigation)
Where are my buddies now?
Who has geotemporal routines similar to mine?
Whos close to me now who shares my interests?
Whos in this place now among my buddies?
What kinds of things happen in this place?
Who can I talk to in this place?
Who (else) comes to this place a lot?
26
People-Centric / Absolute User Location
27
ULocate
28
Active Campus Explorer Buddies
29
People-Centric / Proximity (P2P)
Social Net Hocman
30
Lovegety

• Male/female versions
• Three modes talk, kaoroke, get2
• 5 meter range
• Units beep, blink when opposite sex detected with
  same settings
• Also flash and make different beep for near
  misses

Lovegety (Erfolg)
31
Lovegety Tales from the FieldWired News, 1998

• Millions sold to teens and adults
• Broad demographic can be problematic
• Matches suggested between adults and young teens
• Workarounds developed
• Erfolg website for pre-arranging encounters
• Counter to original spirit of device

32
Social Net

• Implicit interest matching application
• Uses patterns of collocation over time to infer
  potentially shared interests without identifying
  those interests
• Example attend the same gym after work
• When two people suspected to share interests,
  mutual friend sought by system
• Mutual friend receives suggestion to introduce
  two friends
• Visualization of process follows

33
Social Net in Action
B
B Introduce A and C
A
C

• B knows A and C

• A and C dont know each other, but frequently
  collocated

• Social Net recognizes frequent collocation,
  generates a message for B suggesting that B
  introduce A and C

34
Hocman
35
Place-centric / use of physical spaces by people
36
Active Campus Explorer Buddies Map
37
Place Centric / Matching Virtual Places
38
Active Campus Graffiti
Graffiti Associating virtual information with
physical places
Messaging your buddies in a place
39
GeoNotes
40
comMotion

• Using GPS, system notices recurring outages
  guesses these are buildings you habitually go
  into
• Prompts you for labels
• Once places are labeled, you can associate to-do
  items with places
• Then get visual / auditory reminders when youre
  in the place

41
Issues/Challenges

• Getting location
• GPS WiFi Active Badges in smart environments
  etc.
• Mapping location onto places
• How do people conceive of places?
• Overlap, inclusions, ..
• Privacy vs. awareness disclosing location
  information at different levels of granularity
• Design of interfaces for handheld devices
• Methodological how to gather user/usage data
• Trust, accountability, reputation

42
My interests

• Understanding the role of semantic places in
  mobile information services
• Empirical study of how places influence need for
  and willingness to share information
• Systems designed based on findings
• People-centered location-aware systems, using
  absolute location
• Place-centered systems perhaps spanning both of
  Jones and Grandhis categories
• Use of this technology to build stronger
  interpersonal bonds and tighter knit communities

43
Resources and Status

• Grant (with S. Shekhar, J. Riedl, and J. Konstan)
  to buy handheld devices
• Currently, a small set of
• Loaded IPAQs (built-in bluetooth, WiFi, GPS
  cards, cell modem cards, other accessories)
• Digital cameras with Bluetooth, GPS, WiFi
• GPS-enabled cell phones
• Tablet PCs with Bluetooth, WiFi, GPS cards
• Money to buy lots more

44
Status Semantic Places project

• With Quentin Jones Steve Whittaker
• Conceptual framework developed (Jones Grandhi)
• Ongoing interviews investigating peoples
• Concepts of place
• Relationships between places
• Information needs associated with places
• Willingness to share info with others in specific
  place types
• Next Experience Sampling studies
• And after that design, prototyping, deployment,
  field studies

45
Status new work

• People-centered location awareness
• Analysis and matching of geotemporal trails
• Favor coordination systems (Pam Ludford)

46
People-centered location awareness

• Status in an awareness tool includes location
• Location computed by system, incl. mapping to
  semantic places at different levels of resolution
• People can reveal information at different levels
  of resolution to different buddies

47
Scenario

• Robin is a professor at UMN
• She wants to let all the members of her
  department know whether shes in her office or
  elsewhere on campus.
• Shell let the members of her research group know
  her exact location as long as shes on campus.
• She wants students in her class to only know
  whether shes in her office or not.
• Shell let colleagues at other institutions know
  whether shes in Minneapolis, or, if shes
  traveling, what other city shes in
• Finally, shell let her family know her exact
  location at all times.

48
Challenges

• Getting reliable location data
• GPS
• WiFi
• Activity on a fixed computer, other sensors
• Language to allow flexible, yet
  privacy-preserving access policies
• Interface to make it easy to state policies and
  monitor their effect, i.e., see what information
  you are (or would be) revealing to specific
  buddies

49
Geotemporal analysis matching

• Recurring patterns as opportunities for
• Labeling important places individual or group
  (ala ESP Game)
• Introducing people find me a carpool buddy
• Parallel to SocialNet, but
• With absolute locations, not proximity
• Temporal, not just spatial matching
• Perhaps more sophisticated matching

50
Favor Coordination

• Scenario 1 errand coordination
• Scenario 2 just-in-time car pooling

51
Favor Coordination Motivation

• Potential to increase social capital by 1.
  giving people a reason to interact
• 2. building mutual trust
• Potential to save natural resources by enabling
  car-pooling

52
Technical Issues in Favor Coordination

• Assuring only one person fulfills a favor
• Negotiating, displaying locations for
  just-in-time car pooling
• Conserving storage space on cell phone
• Assuring privacy

53
Trust in Favor Coordination

• Trade-off between risk and benefit
• Encourage networks that are likely to foster
  trust and favor trading
• e.g. extended family vs. co-located
• Limited space to display information related to
  trust

54
Reciprocity also a social issue to consider
55
THE END
56
Systems
Social Net Hocman
57
Cool, but

• Can we apply a user-centered approach to create
  novel designs with novel technologies?
• How do we learn about user requirements?
• Lets take an example

58
Place-centered information

• Deliver information to users (on handheld /
  wearable devices) that is relevant to their
  current place (e.g., office, lab, classroom,
  home, grocery store, night club, coffee shop,
  stadium, post office, school, etc.)

59
What wed like to know

• Common places a representative group of people
  are in over the course of a typical day
• Common transitions between types of places
• Information needs relative to a given type of
  place
• Willingness to share information relevant to a
  given type of place

60
Hypothetical examples

• While Im in the grocery store, Id like to know
  if any of the items I typically buy are on sale
• While Im dropping my kids off at school, I
  realized it would be really convenient to find
  another parent at the school to carpool with
• Id like somebody whos passing by the Starbucks
  at Stadium Village to pick me up a Venti Decaf
  Latte and bring it to my office
• I often go from my office to the Rec Center to
  the coffee shop, then back to my office

61
Time Diary (Rieman 1993)
62
Limits of self-reported data

• Basic problem hard for people to remember and
  take time to enter data
• Failure of recall esp. for unremarkable
  everyday events

63
Another approach

• Experience Sampling Method (ESM)
• Subjects are periodically prompted to enter data,
  typically by answering a few questions
• Fixed interval random interval on event (up to
  user)
• Traditionally, prompting was via a beeper and
  questionnaire was filled out with paper and
  pencil
• New technologies
• Call cell-phone, listen to voice prompts, press
  buttons or speak responses
• Blackberry pager
• Handheld computer

64
ESM Tool Tailored for Ubicomp

• Intille et al, Ubicomp 2003
• In addition to traditional prompting modes,
  context-aware triggers
• User is in a particular location
• Users heart rate changes
• A conversation is (or is not) going on
• Ubiquitous, imaged-based sampling
• Audio, video images captured
• Users consult the images when its convenient and
  answer the survey questions using images as
  recall aids

65
Context-sensitive ESM
66
Limitations/Issues inExisting Applications

• Often circumvent established social processes,
  norms
• Technology attempts to mediate interaction
• When used, profiles are poor characterizations of
  people
• Privacy concerns, trust, accountability
• What information is being transmitted to whom?
• Who is this stranger? Introductions are awkward!
• Scaling sometimes problematic
• An application design issue, not technological
• What if everyone had one? and interests arent
  enough!
• Challenges for interaction styles
• Always on, with person
• Interruption

67
Active Badge

• Lets someone be located within a building
• Badges emit IR signals
• Sensors in the environment pick up the signals,
  pass them to a central network

68
Olivetti Active Badge, circa 1990
69
Placing Information in the World

• GeoNotes
• Associate notes (e.g., recommendations, ratings)
  with locations for personal or group use
• E-Graffiti
• Campus use
• comMotion
• Associates to do list items with locations
  individual use
• ActiveCampus
• Big ongoing effort at UCSD expanding to other
  sites