Know Your Network
Do you have questions or comments about this model? Ask them here! (You'll first need to log in.)
WHAT IS IT?
This model allows you to import social network data, and explore how a rumor might spread on it.
HOW IT WORKS
The rules are simple: you want to spread a rumor as fast as possible to everyone in your social network. To get the rumor started, you get to 'whiper' the rumor to five people. For every 'tick', they then have a 50% chance of telling each person they know.
The faster the rumor spreads, the more points you get. Specifically, the payoff you get for telling people drops by 10% for every tick. So make sure you initially whisper the rumor to people who, because of their location in your network, can spread the rumor as fast as possible!
HOW TO USE IT
Press LOAD-GRAPH first to import your data.
The widgets near the "Visualize it!" heading allow you to visually explore your network. You can set the size of nodes based on their various centralities, and you can show labels for those nodes that are in the highest percet, set by the LABEL-THRESHOLD slider. If you want to see a prettier, layouted version of your network, press the "Layout" button.
Click the SETUP RUMOR MODEL button to start spreading the rumor in your network.
You can infect people in various ways; by their specific name, by having the model tell someone based on their network centrality, and by 'touching' them. If you want to infect people by touching them, you have to click the "BY TOUCH" button" first.
Finally, click the GO button to start spreading the rumor. When everybody has heard the rumor, the model will automatically stop. The plot will show you how many people heard the rumor, and the SCORE monitor will tell you how well you did.
If you want to see a network representation of who told whom, click the SHOW RUMOR NETWORK button. This will show you how the rumor spread from each of the five people that you initialy told.
THINGS TO NOTICE
Running the model more than once may (and probably will) yield different results. This is because there is some randomness in how the rumor spreads.
CREDITS AND REFERENCES
Know Your Network. Arthur Hjorth, David Weintrop, Uri Wilensky 2014.
Comments and Questions
extensions [ nw ] turtles-own [ ;; facebook data id name centrality ;; infection variables infected? heard-from ] links-own [ weight ] to load-graph clear-all ;; Hard-code your path if you don't want to get prompted every time: ;; let filename user-file "/path/to/your-network.graphml" let filename user-file if (filename != false) [ nw:load-graphml filename [ set shape "circle" ] nw:set-context turtles links ] ; let no-of-components length (sentence nw:weak-component-clusters) ; ifelse no-of-components > 0 [let degrees-separated 360 / no-of-components ; let counter 0 ; foreach nw:weak-component-clusters [ ; let biggest-component-size max map [count ?] nw:weak-component-clusters ; if count ? != biggest-component-size [ask turtle-set ? [die]] ; ] ; reset-ticks ; ] ; [ ; user-message "No network was loaded, please try again." ; ] end to remove-most-central ask max-one-of turtles [ count my-links ] [ die ] nw:set-context turtles links end to update-layout layout-spring turtles links 0.2 .1 1 end to toggle-labels ;; if any have a label, turn off all labels ifelse any? turtles with [label != ""] [ ask turtles [set label ""] ] [ let the-number-of-turtles count turtles * (100 - label-threshold) / 100 ask max-n-of the-number-of-turtles turtles [size] [set label name] ] end to calc-centrality if centrality-measure = "random" [ ask turtles [set size random 14 + 1] stop ] if centrality-measure = "reset-size" [ ask turtles [set centrality 1 set size 1] stop ] if centrality-measure = "degree-centrality" [ ask turtles [set centrality count my-links] ask turtles [set size 10 * (normalize centrality min [centrality] of turtles max [centrality] of turtles)] stop ] let the-task (word "set centrality nw:" centrality-measure) ask turtles [run the-task] ask turtles [set size 10 * (normalize centrality min [centrality] of turtles max [centrality] of turtles)] end to setup clear-all-plots ask turtles [ set infected? false set label "" set color white set size 3 ] ask links [ set thickness 1 set color grey ] reset-ticks end ; this normalizes a number to-report normalize [value the-min the-max] set the-min ifelse-value (the-min = 0) [.001] [the-min] let normalized (value - the-min) / (the-max - the-min) report normalized end to kill-smaller-components let the-component [] foreach nw:weak-component-clusters [ if length ? > length the-component [set the-component ?] ] set the-component (turtle-set the-component) show count the-component ask turtles with [not member? self the-component] [die] end ;; turtle procedure to show-centralities show (word "Name: " name ", degree: " count my-links ", betweenness: " nw:betweenness-centrality ", closeness: " nw:closeness-centrality ", eigenvector:" nw:eigenvector-centrality ", page-rank: " nw:page-rank) end ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;; Layouts ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; to redo-layout [ forever? ] repeat ifelse-value forever? [ 1 ] [ 50 ] [ layout-spring turtles links 1 (80 / 1) (1 / 1) display if not forever? [ wait 0.005 ] ] end to layout-once redo-layout false end to spring-forever redo-layout true end
There is only one version of this model, created about 11 years ago by David Weintrop.
Attached files
File | Type | Description | Last updated | |
---|---|---|---|---|
dweintrop.graphml | data | GML Example | about 11 years ago, by David Weintrop | Download |
This model does not have any ancestors.
This model does not have any descendants.