Archives Unleashed Web Archive Hackathon (#hackarchives) presentation by Tom Smyth, Allison Hegel, Alexander Nwala, Patrick Egan, Nick Ruest, Yu Xu, Kelsey Utne, Jonathan Armoza, and Federico Nanni.

1. Tracking Discourse on
Social Media
Archives Unleashed: Web Archive Hackathon
Toronto, Ontario
Team Critical Load Average

2. Two events:
● Charlie Hebdo shooting (Jan 7, 2015)
● Bataclan attack (Nov 13, 2015)
Two social media sites:
● Reddit
● Twitter
TRACKING DISCOURSE ON SOCIAL MEDIA
Four approaches:
● Attention span
● Information flow
● Topic modeling
● Network analysis

3. REDDIT COMMENT:
TWEET:

4. REDDIT DATA
~50M comments a month on Reddit
13M comments the week
following Hebdo shooting
25M comments the week
following Bataclan attack
48,840 comments about
the Hebdo shooting
110,520 comments about
the Bataclan attack

5. 6,964,831 bataclan tweets
(english, nov. 13 - nov. 19)
4,280,030 hebdo tweets
(english, jan. 7 - jan 13)

6. All our data

9. Use ALL available resources

10. 60GB of JSON -> 1GB of txt

11. Attention on social media

12. Longitudinal Analysis (spread of information/misinformation)
(http://www.cs.odu.edu/~anwala/files/temp/archivesUnleashedHackathon/Bataclan_Twitter.html)

13. Longitudinal Analysis (evolution of conversation)
day 1 day 2 day 3 day 4 day 5 day 6 day 7
(http://www.cs.odu.edu/~anwala/files/temp/archivesUnleashedHackathon/Bataclan_Twitter.html)

14. Topic Modeling

15. Topic Modeling

16. NETWORK ANALYSIS:
Word co-occurrence pattern for Charlie Hebdo

17. NETWORK ANALYSIS:
Word co-occurrence pattern for Bataclan

18. FURTHER RESEARCH
● Longer time spans
● Other types of events
● Categorization (hashtags or subreddits)

19. This project brought to you by
Team Critical Load Average:
Alexander Nwala, Old Dominion University
Allison Hegel, UCLA
Federico Nanni, University of Bologna
Jonathan Armoza, NYU
Kelsey Utne, Cornell University
Nick Ruest, York University
Yu Xu, USC