What to expect from Twitter data ?
- Diffusion of information
- Structures of political communities
- Semantic specificity of each community
- Relation between candidates and their community
- Interactions between communities
What to expect from Twitter data ?
$3700$ political figures followed
$66$ hashtags tracked
60 millions Tweets between july 2016 and the presidential election
Create a link between accounts if enough retweets present
Graph between August and October 2016
Position : computed by an algorithm simulating gravitational forces
Color : computed by the Louvain algorithm
Graph between August and October 2016
Diffuse mainly in left-wing communities
Graph between August and October 2016
Diffuse mainly in right-wing communities
Graph between August and October 2016
Diffuse mainly in rigt-wing communities
and accounts without communtiy
Graph between August and October 2016
Diffuse mainly in left-wing communities
and accounts without communtiy
Graph of twitter accounts sharing fake news and debunks from the Decodeurs
Difference of share of fake news
and debunks
Graph between August and December 2016
Nodes : Twitter accounts
Links : enough retweets without modification
Graph 2 weeks before the first round
Graph between first and second round
Alluvial graph of political communities from august 2016 to septembre 2017
Each vertival bar is proportionnal to the size of the community
François Fillon wins the right-wing primary
Perturbation because of the Penelope Gate
Alluvial graph of political communities from august 2016 to septembre 2017