Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

A Networked Take on Influence: what we learn from data


Published on

Analyzing data from social interactions on Twitter and Facebook gives us clues as to how people choose to manage their attention. When intelligently predicted, captured and visualized, we witness emerging patterns. Some swear by a certain time of day. Others spend an inordinate amount of time trying to reach certain users deemed "influencers." But its clear that the networked dynamics at play are much more complicated. While there is no simple recipe for predicting information flows in online social spaces, there's plenty of room for optimization.

Published in: Technology

A Networked Take on Influence: what we learn from data

  1. A Networked Take on Influence Symposium & Workshop on Measuring Influence on Social Media | #Influence12 Social Media Lab, Dalhousie University, Halifax Gilad Lotan | @gilgul
  2. Illustra(on:  Ma.  Taylor  
  3. Sculpting Audiences Alex  Dragulescu  
  4. Sculpting Audiences Alex  Dragulescu  
  5. Influence: The ability to disproportionately affect the spread of information purchasing patterns of peers television viewing behavior Within Context => Wanted Action
  6. “Influence” as an Exposed Metric is Problematic•  Simple quantifiable measure•  Innate nature•  People aren’t always rational – may trust someone despite past transgressions•  Exogenous – may come from outside the observed network•  Based on Context
  7. Receptivity of your Audience Flickr:  mangpages  
  8. f(Topic, Network, Timing, Authority, Trust, Historical Behavior, Network Density, Activity…)
  9. Given tweet(s) => if/when to post
  10. Diurnal Patterns
  11. Daily Routines
  12. Geographic Distribution USA   Germany   Indonesia  
  13. Data Reflects Us
  14. Identifying Spikes49ers:  36,    Saints:  32  
  15. Typical Sports Event Game  ends!  49ers   Game  starts  
  16. TV  Show  Social  Movement   Awards     Ceremony  
  17. Breaking News (zipfian distribution)
  18. Audience & Regional Trends
  19. Trends by CategoriesMeme   Entertainment   Lifestyle   Sports   Poli(cs   Holidays  or  Date   Place  or  Loca(on   #UglyNames   #TheDarkKnightRises   #WhenIGetDrunk   Eli  Manning   Obama   Halloween   Chilean   #truthbomb   JonahHill   #TodayImWearing   #ElClassico   StateOfTheUnion   Thanksgiving   Spain   #TextMessagesFro #IWishIHad   Emma  Stone   mAStalker   #GoLakers   World  War  Web   9/11   Staples  Center   Edward   Scissorhands   The  Boy  is  Mine   ESPNU   Arab  Spring   Pi  Day   Wembley   #ThoughtsWheniLookI nTheMirror   Foo  Fighters   #mydivorcesong   Giants  Won   Michele  Bachmann   Happy  Mothers  Day   Baghdad   #10ThingsIFindA.ra c(ve  
  20. GeoSpan
  21. PersistencePersistence  =  A.en(on  
  22. Volatility (across cities)Vola(lity  =  sum  of  edit  distance  in  consecu(ve  slots  of  the  trending  topics  list    
  23. Audience Attention: Real World EventsHow do audiences of different clients react to real world events?
  24. Networked Audiences
  25. Audience Activity@Pepsi  @AP   Pepsi  Promo(on  @AJEnglish   Whitney  Houston’s   Tragic  Death  Announced  
  26. Shape of an Audience #Kony2012
  27. Noblesville,  IN   Birmingham,  AL  Oklahoma  City   Engelwood,  Dayton,  OH   Pi.sburgh  
  28. #IGoToASchool
  29. Miami
  30. Vancouver
  31. New Jersey
  32. Teen Micro-Celebs
  33. Audience Authorities•  Highly central users often have many followers•  Some people are exceptions: not that many followers, but very central (e.g. Jezebel)•  Most valuable users to engage with? Fewer demands on attention (less popular globally), but well-connected within network.
  34. Information Flows
  35. Bridging by Providing Context
  36. Instead of “Influencer”, think about: •  Deviation from the Norm•  Audience Receptivity•  Network Attributes: shape, centrality, clustering•  Bridges: generating trust•  Context Setting
  37. Thank You Gilad Lotan | @gilgul