As part of a research project, I am tasked with finding the most influential pairs in a weighted social network. That is, the intervention does not assign individuals, but pairs of friends (two nodes that are connected). I'm now looking for a mathematical way of solving this problem. My take was to use a well-established centrality measure (say, group indegree centrality) and then approach it as follows: 1. Find most central player, subset his/her friends, take the most group-central player among them. 2. Repeat with second pair, etc. I'm wondering if this is the best way to approach the problem. Particularly, I wonder if there are centrality measures that assess centrality for pairs (I haven't found any). I would be very grateful for feedback.
Thank you in advance!