• No results found

Network Maps for End Users: Collect, Analyze, Visualize and Communicate Network Insights with Zero Coding

N/A
N/A
Protected

Academic year: 2021

Share "Network Maps for End Users: Collect, Analyze, Visualize and Communicate Network Insights with Zero Coding"

Copied!
54
0
0

Loading.... (view fulltext now)

Full text

(1)

A  project  from  the  Social  Media  Research  Founda8on:  h:p://www.smrfounda8on.org  

Network  Maps  for  

End  Users:    

Collect,  Analyze,  

Visualize  and  

Communicate  

Network  Insights  with  

Zero  Coding  

(2)

About  Me  

Introduc8ons  

 

Marc  A.  Smith  

Chief  Social  Scien8st  

Connected  Ac8on  Consul8ng  Group  

 

[email protected]

 

h:p://www.connectedac8on.net

 

h:p://www.codeplex.com/nodexl

 

h:p://www.twi:er.com/marc_smith

 

h:p://delicious.com/marc_smith/Paper

   

h:p://www.flickr.com/photos/marc_smith

 

h:p://www.facebook.com/marc.smith.sociologist

 

h:p://www.linkedin.com/in/marcasmith

 

h:p://www.slideshare.net/Marc_A_Smith

 

h:p://www.smrfounda8on.org

 

 

(3)
(4)
(5)
(6)

Central  tenet      

–  Social  structure  emerges  from    

–    the  aggregate  of  rela8onships  (8es)    

–    among  members  of  a  popula8on  

Phenomena  of  interest  

–  Emergence  of  cliques  and  clusters    

–    from  pa:erns  of  rela8onships  

–  Centrality  (core),  periphery  (isolates),    

–    betweenness  

Methods  

–  Surveys,  interviews,  observa8ons,     log  file  analysis,  computa8onal     analysis  of  matrices  

 

(Hampton  &Wellman,  1999;  Paolillo,  2001;  Wellman,  2001)  

Source:  Richards,  W.   (1986).  The  NEGOPY   network  analysis   program.  Burnaby,  BC:   Department  of     Communica8on,  Simon   Fraser  University.  pp. 7-­‐16

Social  Network  Theory  

(7)

SNA  101  

•  Node  

–  “actor”  on  which  rela8onships  act;  1-­‐mode  versus  2-­‐mode  networks  

•  Edge  

–  Rela8onship  connec8ng  nodes;  can  be  direc8onal  

•  Cohesive  Sub-­‐Group  

–  Well-­‐connected  group;  clique;  cluster  

•  Key  Metrics  

–  Centrality  (group  or  individual  measure)  

• Number  of  direct  connec8ons  that  individuals  have  with  others  in  the  group  (usually  look  at   incoming  connec8ons  only)  

• Measure  at  the  individual  node  or  group  level  

–  Cohesion  (group  measure)  

• Ease  with  which  a  network  can  connect  

• Aggregate  measure  of  shortest  path  between  each  node  pair  at  network  level  reflects   average  distance  

–  Density  (group  measure)  

• Robustness  of  the  network  

• Number  of  connec8ons  that  exist  in  the  group  out  of  100%  possible    

–  Betweenness  (individual  measure)  

• #  shortest  paths  between  each  node  pair  that  a  node  is  on   • Measure  at  the  individual  node  level  

•  Node  roles  

–  Peripheral  –  below  average  centrality  

–  Central  connector  –  above  average  centrality  

–  Broker  –  above  average  betweenness  

E   D   F   A   C   B   H   G   I   C   D   E   A   B   D   E  

(8)

Email  (and  more)  is    

(9)

PaDerns  are  leE  behind  

(10)

There  are  many  kinds  of  8es….  

(11)

World  Wide  Web  

Each  contains  one  or  more   social  networks  

(12)

Welser,  Howard  T.,  Eric  Gleave,  Danyel  Fisher,  

and  Marc  Smith.  2007.  

Visualizing  the  Signatures  of  Social  Roles  in  

Online  Discussion  Groups

.    

The  

Journal  of  Social  Structure

.  8(2).  

Experts  and  “Answer  People”  

Discussion  starters,  Topic  se:ers  

(13)

Tag  Ecologies  I  

(14)

HUB-­‐AND-­‐SPOKE  OF  DECEIT:  When  Enron  employees  communicated  about  legi8mate   projects,  e-­‐mails  were  reciprocal  and  informa8on  was  shared  widely  (right),  but  

communica8ons  about  an  illicit  project  (les)  reveal  a  sparse  network  with  a  central,   informed  clique  and  isolated  external  players.  

Brandy  Aven,  CMU  

h:p://www.sciencenews.org/view/generic/id/330731/8tle/ Informa8on_flow_can_reveal_dirty_deeds  

(15)

Goal:  Make  SNA  easier  

Exis8ng  Social  Network  Tools  are  challenging  

for  many  novice  users  

Tools  like  Excel  are  widely  used  

Leveraging  a  spreadsheet  as  a  host  for  SNA  

lowers  barriers  to  network  data  analysis  and  

display  

(16)

Who  we  are  

People   Disciplines   InsGtuGons  

University  

Faculty   Computer  Science   University  of  Maryland   Students   HCI,  CSCW   Oxford  Internet  Ins8tute  

Industry     Machine  Learning   Stanford  University   Independent     Informa8on  

Visualiza8on   Microsos  Research   Researchers     UI/UX   Illinois  Ins8tute  of  

Technology   Developers     Social  Science/Sociology   Connected  Ac8on  

Network  Analysis     Cornell  

(17)

What  we  are  trying  to  do:  

Open  Tools,  Open  Data,  Open  Scholarship  

Build  the  “

Firefox  of  GraphML

”  –  open  tools  for  

collec8ng  and  visualizing  social  media  data  

• 

Connect  users  to  network  analysis  –  make    

network  charts  as  easy  as  making  a  pie  chart  

Connect  researchers  to  social  media  data  sources  

• 

Archive:  Be  the  “Allen  Very  Large  Telescope  

Array”  for  Social  Media  data  –  coordinate  and  

aggregate  the  results  of  many  user’s  data  

collec8on  and  analysis  

Create  open  access  research  papers  &  findings  

Make  “collec3ons  of  connec3ons”  easy  for  users  

(18)

What  we  have  done:  

Open  Tools  

NodeXL  

Data  providers  (“spigots”)  

ThreadMill  Message  Board  

Exchange  Enterprise  Email  

Voson  Hyperlink  

SharePoint  

Facebook  

Twi:er  

YouTube  

Flickr  

(19)

What  we  have  done:  

Open  Data  

NodeXLGraphGallery.org  

User  generated  collec8on  

of  network  graphs,  

datasets  and  annota8ons  

Collec8ve  repository  for  

the  research  community  

Published  collec8ons  of  

data  from  a  range  of  social  

media  data  sources  to  help  

students  and  researchers  

connect  with  data  of  

(20)
(21)
(22)

Social  Media  Research  Founda8on

 

(23)

He ath er  has  h igh   betw ee nn ess  

NodeXL  

Network  Overview  Discovery  and  ExploraGon  add-­‐in  for  Excel  2007/2010  

A  minimal  network  can   illustrate  the  ways  different   loca8ons  have  different  values  

(24)
(25)
(26)
(27)

h:p://www.connectedac8on.net/2010/04/25/bernie-­‐hogans-­‐facebook-­‐social-­‐network-­‐data-­‐ provider-­‐and-­‐visualiza8on-­‐toolkit/  

(28)

Network  of  connec8ons  among  the  people  who  tweeted  the  term     “PAWCON”  on  19  October  2011  

(29)
(30)
(31)
(32)
(33)
(34)
(35)
(36)
(37)
(38)
(39)
(40)
(41)
(42)

Analogy:  Clusters  Are  Occluded  

Hard  to  count  nodes,  clusters  

(43)
(44)
(45)
(46)

Social  networks  in  TwiDer  among  people  with   at  least  one  connecGon  to  someone  else  who   Tweeted  “Obama”  on  January  25,  2011  

(47)

Network  of  word  pairs  frequently  men8ons  among  people  who  Tweeted  the  name  “Obama”  on   January  25,  2011    

(48)
(49)
(50)

What  we  want  to  do:    

(

Build  the  tools  to)  map  the  social  web  

• 

Move  NodeXL  to  the  web:  

Node  for  Google  Doc  Spreadsheets!  

WebGL  Canvas  

• 

Connect  to  more  data  sources  of  interest:  

RDF,  MediaWikis,  Gmail,  NYT,  Cita8on  Networks  

• 

Solve  hard  network  manipula8on  UI  problems:  

Modal  transform,  Time  series,  Automated  layouts  

• 

Grow  and  maintain  archives  of  social  media  network  data  sets  for  

research  use.  

• 

Improve  network  science  educa8on:  

Workshops  on  social  media  network  analysis  

Live  lectures  and  presenta8ons  

(51)

Work  Items  

Autofill  Group  A:ribute  

Merge  Edges  by  A:ribute  

Modal  Transform  

Merge  Workbooks  

Automated  Dynamic  Filters:  Time  Series  Analysis,  contrast  

Cap8ons  and  Legends  

Upload  to  Graph  Gallery++:  cap8ons,  workbook  

Graph  Gallery++    

User  Accounts,  Repor8ng,  RSS  Feeds,    

Network  Visualiza8on  Web  Canvas  

Import:  RDF,  Wiki,  SharePoint,  Keyword  networks  from  text  

Metrics:  Triad  Census  

Layouts:    

Force  Atlas  2,  Lin  Log,  “Bakshy  Plots”,  Quality  Measures  

Query-­‐by-­‐example  search  for  network  structures  

(52)

How  you  can  help  

Sponsor  a  feature  

Sponsor  Webshop  2012  

Sponsor  a  student  

Schedule  training  

Sponsor  the  founda8on  

Donate  your  money,  code,  computa8on,  storage,  

bandwidth,  data  or  employee’s  8me  

Help  promote  the  work  of  the  Social  Media  

(53)

Contact:  

 

 

Marc  A.  Smith  

Chief  Social  Scien8st  

Connected  Ac8on  Consul8ng  Group  

 

[email protected]

 

h:p://www.connectedac8on.net

 

h:p://www.codeplex.com/nodexl

 

h:p://www.twi:er.com/marc_smith

 

h:p://delicious.com/marc_smith/Paper

   

h:p://www.flickr.com/photos/marc_smith

 

h:p://www.facebook.com/marc.smith.sociologist

 

h:p://www.linkedin.com/in/marcasmith

 

h:p://www.slideshare.net/Marc_A_Smith

 

h:p://www.smrfounda8on.org

   

(54)

A  project  from  the  Social  Media  Research  Founda8on:  h:p://www.smrfounda8on.org  

Network  Maps  for  

End  Users:    

Collect,  Analyze,  

Visualize  and  

Communicate  

Network  Insights  with  

Zero  Coding  

References

Related documents

In the next section, several trade-offs in which RFID technology can be used while providing the necessary security and privacy protection will be addressed.. A common

We show that for the case in which the outflow fault is up-dip from the point of injection, there is a critical injection rate above which the injected fluid floods the full depth of

Bugis is the biggest tribe in South Sulawesi that occupy some regencies like Bone, Soppeng, Wajo, Barru, Sidrap, Pangkep, Parepare, Sinjai, and Bulukumba. Bugis is tribe

This study was conducted to investigate the demographic characteristics of hydatid cyst surgeries in hospitals of East Azerbaijan Province, Northwest of Iran.. Methods:

Originally 13 attributes involved in prediction of heart disease, proposed enhanced prediction of heart disease with feature subset selection using genetic

At forty-one years on the throne, Asa reigned longer, by a year, than either David or Solomon and much longer than Jeroboam, Rehoboam, Abijah, Nadab and Baasha.. Asa reigned

Death is never a very pleasant subject and perhaps it is our reluctance to discuss it that causes so many of the miscon-ceptions about what happens when we die. For the past

The growing popularity of E-Learning has introduced new terms to education, such as Virtual Classroom, where the students and the teacher meet in a virtual classroom. In