Capabilities of the Zooniverse

My talk at #zoocon 2013 at the Adler Planetarium in Chicago.

Transcript

1. Capabilities of the Zooniverse Arfon Smith
2. What we can do.
3. What we’ll be able to do soon.
4. What we’d like to be able to do.
5. 1. Domain model 2. Technologies/tools 3. Data in/out 4. Intelligent systems
6. 1. Domain model
8. Ancient Lives 07.2011 WhaleFM 11.2011 SETILive 02.2012 Snapshot Serengeti 12.2012 Galaxy Zoo 09.2012 Cyclone Center 09.2012 NEEMO 10.2011 Ice Hunters 08.2011 What’s the Score 05.2012 740,000 Total Volunteers 500,000 Total Volunteers Cell Slider 10.2012 Bat Detective 09.2012 Andromeda Project 12.2012 Seafloor Explorer 09.2012 Old Weather 10.2010 The Milky Way Project 12.2010 Galaxy Zoo 7.2007 SPACE CLIMATE HUMANITIES NATURE HEALTH Mergers 11.2009 Moon Zoo 05.2010 Supernovae 03.2010 Planet Hunters 12.2010 Hubble 04.2010 250,000 Total Volunteers Solar Stormwatch 02.2010 THE ZOONIVERSE 2007-2012
9. Shared vocabulary
10. SUBJECT
11. TASK
12. USER
13. TASK SUBJECT USER CLASSIFICATION
18. TASK SUBJECT USER CLASSIFICATION
19. 2. Technologies/tools
20. 1.We picked your project because we thought it would be hard/different.
21. 1.We picked your project because we thought it would be hard/different. (and we think you’re awesome)
26. 2.If you want to send a million people to our website we’re up for it.
44. 3.We’re committed to helping you do the very best science.
45. Ouroboros
46. USER 1 USER 2 USER 3 SUBJECT ALLOCATOR TASK TASK TASK CLASSIFICATION API
47. SUBJECT
48. TASK SUBJECT USER CLASSIFICATION
50. SUBJECT ALLOCATOR
54. USER 1 USER 2 USER 3 SUBJECT ALLOCATOR TASK TASK TASK CLASSIFICATION API
55. CLASSIFICATION API
58. Talk
61. Discuss
63. Tag
65. Collect
68. 3. Data in/out
70. Metadata
73. Data reduction
80. 4. Intelligent systems
81. USER 1 USER 2 USER 3 SUBJECT ALLOCATOR TASK TASK TASK CLASSIFICATION API
82. USER 1 USER 2 USER 3 SUBJECT ALLOCATOR TASK TASK TASK CLASSIFICATION API
83. USER 1 USER 2 USER 3 SUBJECT ALLOCATOR TASK TASK TASK CLASSIFICATION API
84. USER 1 USER 2 USER 3 SMART STUFF SUBJECT ALLOCATOR TASK TASK TASK CLASSIFICATION API
85. Measure everything
88. 0 0.2 0.4 0.6 0.8 1 1.2 0 1 2 3 4 5 6 7 8 fraction of users tutorial step no Planet Hunters Snapshot Serengeti Cyclone Center
89. Decisions informed by data
90. A B C
91. 830,000 volunteers
92. 300,000,000 analyses
93. Combining Human and Machine Intelligence in Large-scale Crowdsourcing Kamar, Hacker & Horvitz - http://is.gd/horvitz
94. Dynamic Bayesian Combination of Multiple Imperfect Classifiers Simpson et al. - arXiv:1206.1831
95. SUBJECT USER MS MU
96. When is a Subject complete?
97. Who is the best person to look at this Subject?
98. How likely is someone to leave the site in the next five minutes?
99. What task is this person best at?
100. How well is this person performing compared to earlier sessions?
103. Ancient Lives 07.2011 WhaleFM 11.2011 SETILive 02.2012 Snapshot Serengeti 12.2012 Galaxy Zoo 09.2012 Cyclone Center 09.2012 NEEMO 10.2011 Ice Hunters 08.2011 What’s the Score 05.2012 740,000 Total Volunteers 500,000 Total Volunteers Cell Slider 10.2012 Bat Detective 09.2012 Andromeda Project 12.2012 Seafloor Explorer 09.2012 Old Weather 10.2010 The Milky Way Project 12.2010 Galaxy Zoo 7.2007 SPACE CLIMATE HUMANITIES NATURE HEALTH Mergers 11.2009 Moon Zoo 05.2010 Supernovae 03.2010 Planet Hunters 12.2010 Hubble 04.2010 250,000 Total Volunteers Solar Stormwatch 02.2010 THE ZOONIVERSE 2007-2012
104. Thanks arfon@zooniverse.org