How to calculate the weight of an inferred edge based on the existing-edge weights

Hi, all.

Suppose I have this:


(a)-[:LIKES {weight:5}]->(b)
(a)-[:LIKES {weight:13}]->(c)

And now I want to infer the edge (b)-[:SIMILAR_TO {weight?}]->(c).

That is, I want infer that two books are similar because the same reader liked both of them. The reader liked each book with a different intensity (LIKES(weight)), and I want these weights to influence how similar I think the books are (SIMILAR_TO(weight)).

My question: are there any best practices for inferring SIMILAR_WEIGHT(weight) from the two LIKES(weight) values?




The following article might provide some insights:

Also this about triadic closure:

Thanks, Niklas, I'll look.