Talk given at the CSCW 2013 Workshop on Social Media Question Asking, February 2013

1. Using Public Social Media to
Find Answers to Questions
Jeffrey Nichols
Workshop on Social Media Question Asking - CSCW 2013
jwnichols@us.ibm.com
@jwnichls

2. Asking questions on social media…
What’s a good digital camera?

3. Can we leverage the large amount of data
created by public social networks?

4. The Information Iceberg
Information revealed through status updates
Use the information we have
to get the information we don’t
Useful information known to members of social network

5. Examples

6. Where might this be helpful?
• Questions about an event that are best
answered soon after the event
• Questions for which there might be a diversity of
opinion
• More?

7. How feasible is this approach?
• Will people answer questions from strangers?
• Will use of an incentive increase responses?
• What is the quality of the answers?

8. Concrete Prototype: TSA Tracker
Crowdsourcing airport security wait times through Twitter
Step #1. Watch for people
http://tsatracker.org/
tweeting about being @tsatracker , @tsatracking
in airport
Step #2. Ask nicely if they
would share wait time to
help others
Step #3. Collect responses
and share relevant data
on web site
Step #4. Say thank you!
Key Question:
Will people respond to questions
from strangers?
8

9. Questions
From @tsatracker (includes incentive)
“If you went through security at <airport code>, can you
reply with your wait time? Info will be used to help other
travelers”
From @tsatracking (no incentive)
“If you went through security at <airport code>, can you
reply with your wait time?”

10. Concrete Prototype: Product Reviews
Step 1. Identify owners of a product
Key Questions:
Step 2. Ask focused question about product Will people respond to questions
in this different domain?
• How is the image quality?
• Does it take good low light pictures? Will people respond to follow-up
• How quickly does it take a picture after pressing questions at the same rate?
the shutter button?
• How durable is it?
Do responses contain useful &
• What accessories are must haves? accurate information?
• Etc…
Step 3-4. Ask more questions if user responds
Step 5. Visualize results as structured product review (future work)

11. Results…

12. Suspended!
• @tsatracking account (no incentive condition)
given 1 week suspension after asking 150
questions
• Did not violate Twitter Terms of Use
• Exceeded threshold for blocks or message
marked as spam
• Neither of our other accounts were
suspended

13. Results
Key Question:
Will people respond to
questions from strangers?
Answer:
42% response rate
44% of answers
received in 30 mins
No significant difference
between any conditions
(taking into account suspension)

14. Follow-up Question Results
• Significant differences between all 4 questions (H=50.12, df=3, p < 0.0001, Kruskal-Wallis)
and just the 3 follow-ups (H=25.46, df=2, p < 0.0001, Kruskal-Wallis)

15. Response Quality (Coding)
Off-topicInfo per
Relevant Answer
Average Info per
Response Count
Multi- Message
But Useful Info
Wrong Answer
Response
Response
Response
Overall
Breakdown
Tablet 258 71% 19% 3% 1.82 0.48
Food Truck 111 82% 6% 6% 1.69 0.46
Thinks
# Irrelevant No Didn't know we're
Irrelevant Responses Experience or understand a bot
Response Tablet 75 63% 11% 7%
Breakdown Food Truck 20 25% 30% 0%
For more, come see talk on Tuesday, 11am in Regency West 5

16. Qualitative Results
• @tsatracker account picked up 16
followers
• Many positive responses (“this will be
great for travelers”)
• Only one slightly negative response (“this
is creepy”), but that person also gave an
answer

17. What’s next?
Can we build technology to support this
process?
• What do we need to know about potential
answerers to better target questions?
• Can we infer who will answer and who will not?
Feasibility in other domains?
• Influence, marketing, etc.?

18. Engagement Continuum
qCrowd
manual assisted automatic
Send this:
Send
Humans do all the work Analytics streamline decisions: System-driven engagement
“press button to engage”
• Keyword filtering • Scenario-based filtering • Rule-based engagement
• Unstructured engagement • Smart engagement recommendations • Exception identification
• Domain-independent analytics (e.g., based on location inference) and notification
• Customizable engagement scenarios • Intelligent transition to
• Domain-specific analytics human-driven engagement
as desired
18

19. User Modeling to Aid Engagement
• Create an OmniProfile of
each customer from social
media Omni
• Get to know each customer
Profile
as a unique individual
• Employ OmniProfiles to external traits intrinsic traits
better target messages Transaction history Personality
• Try to ensure that only Web interaction Willingness
Demographics Social relationships
those who are willing are … …
contacted

20. Modeling and Deriving Personality
Map the use of words, frequency, &
correlation with Big5 based on LIWC
“Agreeableness”
wonderful (0.28), together (0.26) …
porn (-0.25), cost (-0.23)
Openness
Conscientiousness
Extraversion
Agreeableness
Neuroticism
[Tausczik&Pennebaker 2010, Yarkoni 2010]

22. To wrap up…
• Interaction on social media enables a variety of
applications
• Data collection through real-time targeted
question asking is a potentially useful
application
• Collecting information using this approach is
feasible and produces quality information
• We have built algorithms and technology to
facilitate this approach

23. Thanks!
For more information, contact:
Jeffrey Nichols
jwnichols@us.ibm.com