1 of 33

Visualizing Targeted Audiences

Saiph Savage, Carlos Toxtli, Angus Forbes, Grant McKenzie, Shloka Desai, Tobias Hollerer

2 of 33

Traditional Media

-Only select persons are content producers.

-One-way non-reciprocal communication.

-Hard to measure audience’s reactions and interests.

3 of 33

Social Media

-Every internet user can be a content producer

-Audience can participate and interact

-Easily measure audience’s reactions and traits

4 of 33

What is an Audience?

“Group of people who encounter a work of art, literature, theater, music, video games, or content in any medium.”

-The people formerly known as the audience, Rosen J. 2012

5 of 33

Motivations

What is important when people target their audiences?

Savage et al. identified 3 key aspects:

  • People want to target audiences whose interests match what they are sharing.
  • People use their audiences to help distribute content.
  • People want to be relevant to the audience they target; includes sharing things that are relevant spatially.

Data visualizations could help users easily gauge the traits of their audience.

6 of 33

Motivations

  • Long lists make it �difficult to gauge �the overall traits of an audience.
  • Not Clear what’s the best way to engage with my friends who love Lady Gaga! �What other things do they have in common?

7 of 33

Motivations

Filters may limit understanding the diversity of an audience.

8 of 33

Related Work

Expert Search�SaNDVis: A tool for finding experts in a social network based on expertise on a certain topic, and their locarion.�Perer et al. VAST 11’�People Recommendation Algorithm :�-Recommending People on Social Networking Sites� Chen et al, CHI 09’�-Recommending targeted strangers from whom to solicit information on social media� Mahmud et al IUI 13’

Not focused for the particular task of targeting audiences.

Needs of User might be different!

SaNDVis!

9 of 33

What we propose

  • On the fly find people relevant to users sharing needs.
  • Provide data visualizations that give users overviews and details of their targeted audience.�Considering Interests, Location, and Social Connections.
  • Study how data visualizations change (if any) how people select their audiences

10 of 33

Workflow

Interest Detection

People’s Profiles

User Modeling

}

Person’s Search

Query

Recommendation Engine

Visualization Engine

Input

11 of 33

Components

12 of 33

Hax: Understanding your audience’s interests

13 of 33

Visualizations http://socialquery.ml

14 of 33

Transparent Interface

15 of 33

Social Spread Interface

16 of 33

Social Spread Interface

17 of 33

Location Based Interface

18 of 33

Usability Inspection

19 of 33

Procedure

  • 15 users from the Facebook group.
  • Conducted 15 targeted sharing tasks, and had to find 10 candidates for each task.
  • Each task belonged to one of the following scenarios: �-Find audiences to collaborate in an interest online�-Find audiences to collaborate in an interest offline�-Find audiences to help get online participation�-Find audiences to help get offline participation
  • Participants completed a survey about their experiences with Hax. We used Ground Theory to code the responses.

20 of 33

Experiences from using Data Visualizations to Target Audiences

  • Serendipitous Discoveries
  • Visualizing Diffusion and Participation
  • Audience Diversity
  • Audience Verification

21 of 33

Lessons Learned

  • Data visualizations empower users to consider spatial constraints, have cultural sensitivity and diversity.
  • Need to design tools that bring diversity to users

22 of 33

Discussion

  • The Internet and social media in particular are changing how people interact with their audiences.
  • We need to design communication tools that consider the new relationships emerging among people and their audiences.

23 of 33

24 of 33

Backup Slides

25 of 33

Final Thoughts

  • What are the nature of the messages that incite the most reactions online and offline?
  • How can we examine whether online interactions can translate into offline collective action?

26 of 33

Challenges

  • Information Overload
  • Audiences and Authors can turn destructive
  • Unclear relationships between audiences and authors
  • Hard to design relevant tools

27 of 33

Designing Tools for Targeting Audiences

28 of 33

Recommendation Engine

1. Users interests discovered.

2. Given a query find its relevant

topic and relevant users

“Animal Rights”

Content

Augmentation

Topic Modeling Module

Pets

29 of 33

Visualization Engine

Transparent

Social Spread

LBS

30 of 33

Social Spread Interface

31 of 33

Recommendation Engine

Topic Modeling Module

Users’ Profiles

Geography

Pets

Politics

32 of 33

Motivations

  • Audience members that feel signaled out are more likely to participate.
  • Users want to understand their audiences to better engage with them.
  • Data visualizations could help users easily gauge the traits of their audience.

33 of 33

Comments

Comments