The algorithm they describe is apparently called "EdgeRank." All I can find via Google are really bad infographics and worse layman explanations (ones that make it clear the writer doesn't understand basic graph theory, or is adulterating their understanding for whatever reason).
Can anyone post a link to a research-level paper explaining the algorithm? The name EdgeRank makes me think it's a ripoff of PageRank, so are they solving some sort of eigensystem for, what, an incidence matrix instead of an adjacency matrix?
Ah. It appears to be nothing more complicated than computing this linear formula given, and Facebook has a thousand tiny little knobs for adjusting the parameters.
Maybe the problem is that their algorithm is not really an algorithm. PageRank was good because it had an interpretation as a random walk that made sense (before SEO). EdgeRank can only be interpreted as an arbitrary formula.
Can anyone post a link to a research-level paper explaining the algorithm? The name EdgeRank makes me think it's a ripoff of PageRank, so are they solving some sort of eigensystem for, what, an incidence matrix instead of an adjacency matrix?