Hacker News new | ask | show | jobs
by JohnKemeny 635 days ago
Degree, closeness, and betweenness all seem like poor choices to make such an important decision. Why not HITS?
2 comments

You can't have a correct vision of Italian music (staying only on that subsubsubcategory) of the 80s, 90s, 00s, 10s and 20s like that, at very best you'd end up ignoring significant artists and whole genres.
HITS doesn't refer to musical hit records. It's a network analysis technique.

https://en.wikipedia.org/wiki/HITS_algorithm

Wow it's the first time that I don't know part of what i'm talking about and i'm still right...

MySpace is mostly erased (prominent Italian musicians for some music waves of the 00s and 10s were there), some sites are erased, what pertains to radio and television is mostly not indexed on the public internet and there's significant digital divide between genres. In the Italian context, you'd end with a weird pantheon of rock, rap and popstars and not much else left, in conclusion you would... expunge significant artists and whole genres from results.

I'm not certain HITS could capture what appears to be the temporal component of the network the author incorporates.
I agree that it's difficult, I don't know much about centrality measures for dynamic networks.

However, degree sounds like a bad measure, since the value can be drastically increased by participating in a (few) big project(s).

All correct, it was also one point raised by the reviewers. They had some suggestions about temporal centrality, but in the end I didn't followup on that because this centrality analysis was just a descriptive part that didn't really support any of the primary nor secondary claims in the paper.