Question about the working precision for the pagerank algorithm


(Ddu1) #1

I am trying out the pagerank which is in neo4j-graph-algorithms. My question is very specific:

Is there a way to set the working precision for the algorithm execution?

I am asking because I am comparing results from neo4j-algorithm with those from other implementations (networkx, and also my own through scipy). One thing I noticed is that the implementation in neo4j-algorithm returned a lot of zero scores while other implementations return small but non-zero values. I suspect this is due to the precision cutoff in the neo4j-algorithm implementation. Not sure about what the default cutoff is, but I did a bit of research and did not find any information about this precision customization. This is important because pageranking on large graphs tend to produce small scores. I wonder whether I miss anything here.

Thanks