Take-aways


classify + quantify a slice of the firehose

Social Media Pulse

Virginia quake

Social


~ a decade into the social data revolution [HBR, 2009]:

Twitter, Tumblr, Foursquare, Facebook, Disqus, WordPress, IntenseDebate, GetGlue, StockTwits, Estimize, Newsgator, Bitly, del.icio.us, flickr, Google+, Instagram, Metacafe, Panoramio, Photobucket, Plurk, Reddit, StackOverflow, Vimeo, VK, YouTube, ....


sharing, communicating, networking

cheaper than a census, real-time, many participants...

Last week on Twitter & Tumblr


Twitter / Tumblr firehose data

Focus

social channels span full human space

Ukraine   Golden Globes
          [ NBC.com ]     [ socialitelife.com ]


identify, analyze isolated events

Expected







Unexpected

Expected

warning/forecast + conversation

Winter Storm Janus
Virginia quake

Unexpected

unprecedented simultaneous observation

Expected


Winter Storm Janus
Winter Storm Janus + smooth
Virginia quake
Virginia earthquake

Unexpected

Social Media Pulse


(I promise this is the only slide with equations)


unexpected event at time \(t=0\)


probability of someone sharing, \( p_1(t) = \lambda e^{-\lambda t} \)


\(N\) people \(\rightarrow \sum_{i=1}^{N} \text{Exp} \sim \frac{\beta^{\alpha}}{\Gamma(\alpha)} t^{\alpha - 1} e^{-\beta t}\)


\( \rightarrow \alpha, \beta \)
so what?

divergence






parameters \(\rightarrow\) reality

Virginia quake

news stories begin to circulate
Virginia quake

Time to peak

Half-life

Response time

Activity volume