Investigating Collaborative Mobile Search BehaviorsShahriyar Amini, Vidya Setlur, Ina Xi, Eiji Hayashi, Jason HongCarnegie Mellon University, Nokia Research Center
August 29, 2013
2
Conducted a survey on users’ most recent collaborative mobile search.• 63 Participants
– (35 Male, 26 Female, 2 Skipped)• 68.2% between 18-35 years old• Participants used mobile search frequently
– 19% searched daily– 52% weekly
3
Collaborators often search with more than one device.• 57.1% searches with one other person• 77.8% collaborations are co-located• More than 50% used more than one device• 87.3% share results through talking• Less familiar with the area of search than if
searching alone.
4
5
Conducted a study with search app.• 42 Participants (28 male, 14 female)• 73.8% between 18-25, others: 26-35• 23 students, others: writers, attorneys, etc.• Studied stand-alone app vs. collaborative
version of the app• Searched and agreed on a restaurant where
they would both like to eat lunch• 2 tasks with each version including one practice
6
Users can view and share results.
7
Collaborative features promoted exploration.• Collaborative searches took longer:
– 5.81 vs 7.42 mins (p < 0.01)• Collaborative searches involved more
detailed view pages:– 11.90 vs 18.33 detail page views (p < 0.01)
• Non-collaborative searches resulted in replication of the search process and comparison of the returned results.
8
Collaborators usually exercised two approaches.
9
Exploratory
Targeted
Participants took into account the opinion of those not present.
10
Design Implications and Conclusion
Facilitate communication:Provide an opportunity to explain actions.
Offer collaborative filters/omission lists:Enable users to express dislikes.
Optimize for friends and family:Offer pre-sets and expose preferences.
11
12
Users can view and share results.
13
Users can re-use previous queries.
14
Collaborators used the Picks list most often.• Median 3 picks added• Median 0 picks removed• Participants were very
aware of the notifications• Notifications were used 15
times total by 9 users• Query cloud used 19 times
by 10 participants• More effective with session
histories and popular searches
15