This document describes a study exploring how social interactions can help with individual search tasks. The study involved characterizing everyday and difficult searches through surveys, and observing participants completing search tasks with or without social support. Results suggested social interactions helped with 40-61% of searches, and in observations, participants employed tactics like searching, asking friends/contacts, and targeted questions. The study provided evidence that social interactions may help searchers by providing cognitive benefits like feedback and iteration.
18. Photo Credit: David Wild
Photo Credit: Thomas Hawk
classic view of search revised view of search
activity can involve other people (remotely or
single‐user activity
co‐located)
query‐response mechanism highly dynamic, fluid process
few keywords, short sessions search can last over an extended period
search takes place in a rich ecology of social and
log files from single search engine information resources
19. Photo Credit: David Wild
Photo Credit: Thomas Hawk
classic view of search revised view of search
activity can involve other people (remotely or
single‐user activity
co‐located)
query‐response mechanism highly dynamic, fluid process
few keywords, short sessions search can last over an extended period
search takes place in a rich ecology of social and
log files from single search engine information resources
20. Photo Credit: David Wild
Photo Credit: Thomas Hawk
classic view of search revised view of search
activity can involve other people (remotely or
single‐user activity
co‐located)
query‐response mechanism highly dynamic, fluid process
few keywords, short sessions search can last over an extended period
search takes place in a rich ecology of social and
log files from single search engine information resources
21. With a revised notion of search...
...comes a revised method of study
Searches are composed of:
Activities
Goals
Operators
22. With a revised notion of search...
...comes a revised method of study
Searches are composed of:
Log files reveal operators
Activities
Goals
Operators
23. With a revised notion of search...
...comes a revised method of study
Searches are composed of:
Observations reveal activities
Activities
Goals
Operators
27. Information seeking in physical contexts
Photo Credit: Rachael Lovinger
A LLEN 1977; C ROSS ET AL 2001; C ROSS & S PROULL 2004; B ORGATTI & C ROSS 2003
28. Information seeking in physical contexts
Photo Credit: Rachael Lovinger
A LLEN 1977; C ROSS ET AL 2001; C ROSS & S PROULL 2004; B ORGATTI & C ROSS 2003
29. Information seeking in physical contexts
Photo Credit: Rachael Lovinger
HIERARCHY (STATUS) PROXIMITY SOCIAL OBLIGATIONS
A LLEN 1977; C ROSS ET AL 2001; C ROSS & S PROULL 2004; B ORGATTI & C ROSS 2003
30. (Joint) collaborative search online
M ORRIS 2008; P ICKENS ET AL . 2008; P AUL & M ORRIS 2009; S HAH 2008.
SEARCH TOGETHER CO-SENSE
31. (Joint) collaborative search online
M ORRIS 2008; P ICKENS ET AL . 2008; P AUL & M ORRIS 2009; S HAH 2008.
SEARCH TOGETHER CO-SENSE
36. study one survey: everyday searches
study two survey: difficult or failed searches
study three observaTons: cogniTve benefits of
social interacTons during search
38. Large scale characterization studies
SEARCH STUDY N=
generic or “everyday” 150
difficult or failed 150
Mechanical Turk
Photo Credit: egoldviet (USED WITHOUT PERMISSION)
39. Large scale characterization studies
SEARCH STUDY N=
generic or “everyday” 150
difficult or failed 150
INFORMATIONAL
searching for informaTon
assumed to be present, but
otherwise unknown
40. Large scale characterization studies
SEARCH STUDY N= INFORMATIONAL
generic or “everyday” 150 59%
difficult or failed 150 87%
INFORMATIONAL
searching for informaTon
assumed to be present, but
otherwise unknown
41. Large scale characterization studies
SEARCH STUDY N= INFORMATIONAL
generic or “everyday” 150 59%
difficult or failed 150 87%
BEFORE DURING AFTER
• reflecTon, synthesis
• search preparaTon • search execuTon
• feedback, iteraTon
• problem formulaTon • lookup, foraging, re‐finding
• sensemaking
42. Large scale characterization studies
SEARCH STUDY N= INFORMATIONAL SOCIAL INTERACTIONS
generic or “everyday” 150 59% 40%
difficult or failed 150 87% 61%
BEFORE DURING AFTER
• reflecTon, synthesis
• search preparaTon • search execuTon
• feedback, iteraTon
• problem formulaTon • lookup, foraging, re‐finding
• sensemaking
45. Cognitive benefits of social searching
(study three)
E VANS , K AIRAM , P IROLLI 2009; E VANS , K AIRAM , P IROLLI 2009
46. Recruiting subjects
PRE-TEST SURVEY PARTICIPANTS (N=8)
• knowledge of energy policies
• computer and internet use
• search experTse
• social acTviTes
Icon Credit: Iconaholic.com, dryicon.com E VANS , K AIRAM , P IROLLI 2009; E VANS , K AIRAM , P IROLLI 2009
47. Recruiting subjects
PRE-TEST SURVEY PARTICIPANTS (N=8)
• knowledge of energy policies
• computer and internet use
• search experTse
• social acTviTes
Icon Credit: Iconaholic.com, dryicon.com E VANS , K AIRAM , P IROLLI 2009; E VANS , K AIRAM , P IROLLI 2009
48. Two task questions
Icon Credit: iconfactory.com, http://ecotechdaily.com/wp-content/uploads/2008/05/oil_drums_450.jpg
49. Two task questions
55 mph
“If we lowered the speed limit
nationally to 55 mph, how many
fewer barrels of oil would the
U.S. consume every year?”
Icon Credit: iconfactory.com, http://ecotechdaily.com/wp-content/uploads/2008/05/oil_drums_450.jpg
50. Two task questions
55 mph Pyrolytic oil
“If we lowered the speed limit “What role does pyrolytic oil (or
nationally to 55 mph, how many pyrolysis) play in the debate over
fewer barrels of oil would the carbon emissions?”
U.S. consume every year?”
Icon Credit: iconfactory.com, http://ecotechdaily.com/wp-content/uploads/2008/05/oil_drums_450.jpg
51. Two search conditions
SOCIAL NON-SOCIAL
Talk-aloud protocol
• friends • search engines
(email, phone, IM, etc.) (Google, Yahoo)
• social networks • Wikipedia
• blogs
• QuesTon‐Answer sites
Icon Credit: fasticon.com, deleket.com, sykonist.deviantart.com
52. Two search conditions
SOCIAL NON-SOCIAL
Talk-aloud protocol
• friends • search engines
(email, phone, IM, etc.) (Google, Yahoo)
• social networks • Wikipedia
• blogs
• QuesTon‐Answer sites
Icon Credit: fasticon.com, deleket.com, sykonist.deviantart.com
61. Depth of processing
SCORE DESCRIPTION
1 identified or perceives facts, data, or info
understands the meaning of info;
2
presents a translaTon of info
3 integrates and synthesizes learned info
TIME
Ex. of learning of
one fact over time
SCORE 1 2 3
62. Depth of processing
SCORE DESCRIPTION
1 identified or perceives facts, data, or info
understands the meaning of info;
2
presents a translaTon of info
3 integrates and synthesizes learned info
TIME
Ex. of learning of
one fact over time
(FINAL)
SCORE 1 2 3
SCORE FOR FACT #1
76. Results
Thinking before Thinking a'er
posTng quesTon receiving replies
60% OF USERS 0% OF USERS
NETWORK ASKING TIME
29% OF USERS 100% OF USERS
TARGETED ASKING TIME
77. Results
Thinking before Thinking a'er
posTng quesTon receiving replies
60% OF USERS 0% OF USERS
NETWORK ASKING TIME
29% OF USERS 100% OF USERS
TARGETED ASKING TIME
More thinking before posting
questions on social networking sites
78. Results
Thinking before Thinking a'er
posTng quesTon receiving replies
60% OF USERS 0% OF USERS
NETWORK ASKING TIME
29% OF USERS 100% OF USERS
TARGETED ASKING TIME
More thinking after receiving replies
sent from targeted friends
80. Social networking sites: Thinking before posting to many
“Now, what could I say?”
“Is this better phrased
as two questions?”
“Let’s see...what do I
really want to be asking?”
82. Targeting friends: Thinking after getting replies from few
Pyro means...
instant
messenger
Long reply...
email
83. Targeting friends: Thinking after getting replies from few
“What are people’s
average driving speeds
anyway?”
“Given that, then I need
to also know...”
“If ‘pyro’ means fire, then
this might be a process
to...”
Pyro means...
instant
messenger
Long reply...
email
85. Nature of Replies
NETWORK ASKING TARGETED ASKING
• short, conversa2onal • long, detailed
• funny, not relevant • focused, relevant
“Because no one drives the
speed limit”
“Lots more waste sitting idly
in traffic”
“Isn’t that something I rub
on my [body]? Are you still
in San Francisco?”
86. Nature of Replies
NETWORK ASKING TARGETED ASKING
• short, conversa2onal • long, detailed
• funny, not relevant • focused, relevant
“Because no one drives the “15-25% savings. But if
speed limit” those cars were electric, we’d
have all those barrels left to
use for something else.”
“Lots more waste sitting idly
in traffic”
“There’s no one national speed
limit, there are two: 55 miles
“Isn’t that something I rub
per hour in general, 65 miles
on my [body]? Are you still
per hour for certain roads.”
in San Francisco?”
95. Social Presence Theory
“A communicator’s sense of awareness of
the presence of an interaction partner”
Short, Williams, & Christie 1976
Photo Credit: Carlo Nicora; George Duncan
96. Social Presence Theory
“A communicator’s sense of awareness of
the presence of an interaction partner”
Short, Williams, & Christie 1976
Photo Credit: Carlo Nicora; George Duncan
97. Social Presence Theory
“A communicator’s sense of awareness of
the presence of an interaction partner”
Short, Williams, & Christie 1976
Photo Credit: Carlo Nicora; George Duncan
109. Inhabitedness
Social Presence
+ Structure of the Space
nature of the relaTonship features of the channel
!e strength (e.g., strong !es, weak !es) mul!media content
relaTve group membership the acTviTes supported
social network size “social objects”
apparent idenTty
pseudonym vs. real name
visibility
frequency of updates
110. Next Steps...
Analytical • operaTonalize the model
• develop hypotheses about how
inhabitedness predicts behaviors
Methodological • test our predicTons experimentally
117. Thank you!!
Third year class Supervisor Third year advisor
Kaya de Barbaro David Kirsh Andrea Chiba
Matthew Leonard
Josh Lewis
Anne Marie Piper
Outside Collaborators Friend Helpers!
Ed H. Chi PARC Chris Messina
Peter Pirolli PARC Sharoda Paul
Sanjay Kairam PARC Michael Bernstein
Michael Muller IBM Research
Elizabeth Churchill Yahoo! Research
118. Discussion
Search as an ac2vity:
• Searches are embedded in rich acTviTes that benefit from social
interacTons with others, and from social communiTes online
Research Findings:
• More thinking before asking quesTons to large social networks;
• More thinking amer genng replies from targeted friends
Towards a theory:
• Social presence theory alone doesn’t explain most social search
behaviors
• Inhabitedness may explain some of these behaviors
• This is a rich area for future design work
Photo Credit: Peter Lee