Analyse de réseaux sociaux
Une autre façon de visualiser la liste est par un réseau bipartite ou bimodal : l'analyse de réseaux sociaux permet de relier des "nœuds" (par exemple, dans une correspondance, les auteurs et les destinataires d'un courrier) par des "liens" (la quantité de lettres qu'ils s'envoient) ; ici, le réseau est bimodal : il y a deux types de nœuds : les auteurs (représentés par leur adresse de courrier électronique), et les sujets (représentés par les fils). Les nœuds "auteurs" ne sont pas reliés entre eux : chaque auteur poste dans un sujet donné mais à l'ensemble de la communauté. Le graphe permet de voir quels sujets sont reliés entre eux (par le fait qu'ils ont des auteurs en commun) et quels auteurs sont reliés entre eux (par le fait qu'ils ont des sujets en commun).
thread social network analysis
A bipartite, or bimodal graph is a network graph where there are two different kinds of nodes (here the green nodes represent email addresses, thus people writing to the list, and the red nodes represent the threads, and according to the definition of threads, they also represent the topics). An (undirected and weighted) edge can only exist between red and green dots as it is defined by the number of messages a person writes in that particular topic. The size of the nodes represent the total number of messages (for a green dot, it is the total number of messages that a person writes (here in 2001) and for a red dot it is the total number of messages a topic received.
Many threads consist in one person posting a message, somebody answering, possibly the first person making a summary of private answers he may have received and that's it. A filtering of the smaller dots allow to focus on the most interesting threads, the four (or six) we unveiled in the time series.
16->17
Here the emails are radically anonymised, but the interesting ideas that the graph reveals is the proximity of threads (and topics), regarding actors that share topics in common, and the proximity of actors that share topics in common. The "simmetry change" topic defines an isolated component: it has its own contributors, different from other topics. The actual structure of a component around a topic, or around an actor can also lead to conclusions about the nature of the thread (or contributors): A collection of successive answers of two people is likely to be an argument between them. A thread surrounded by numerous green dots with small edges is likely to be a balanced debate of different and diverse opinions.
17->18
