Interaction Networks

We defined an interaction as the event in which two editors happen to edit under the same section. The rationale behind this was that each section represents an aspect of the projects and if two editors edit under the same section, it implies that they worked for the same aspect of the project. Hence, they interact in this sense. We defined the strength of interaction between two editors as the number of interactions that were recorded for those two editors. It should be noted that a single interaction corresponds to a single common section, irrespective of the number of edits done by the two editors individually in that section. Once we define an interaction between two editors, we can make a weighted and undirected interaction network for the editors of a team.
For the purpose of analyses, we considered only the significant interactions in the network. We compared the interaction network of each team with a null model. In this null model, we conserved the weighted degree of the editors. We compare the weight of each edge against the weight according to hypergeometric distribution, which gives us the probability of co-editing a section given the total number of sections in the whole wiki. A Z-score was computed for the weight of each edge. Edges which had a Z-score less than two were ignored.