4 ms·
As explained above, you could create a graph where the nodes are the users and submissions. Then you could have bidirectional links between each user and every
by amichail 4y ago
As explained above, you could create a graph where the nodes are the users and submissions. Then you could have bidirectional links between each user and every submission made by that user.
You could then compute the importance of each node in the graph using PageRank, thus giving you the importance of each user and the importance of each submission.
- verdverm 4y ago> As explained above, you could create a graph where the nodes are the users and submissions. Then you could have bidirectional links between each user and every submission made by that user. Then you aren't doing the PageRank algo anymore, it's something different. PageRank uses a directed graph, "bidirectional" edges implies an undirected graph
- amichail 4y agoPageRank works fine on bidirectional links. The web has pages that link to pages that link back to them.
- verdverm 4y agotwo directional links is not the same as "bidirectional" links What happens if I create a site which links to every other site on the internet? Should my site rank highly? Probably not To do well with PageRank, you want to be a site which many others link to. It does not matter how many you link to. Direction of links matters
- amichail 4y agoIt's true that cheating the system is easy with bidirectional links. So maybe limit the upvote rate and take into account the time at which the link is made so that early links count for more than later ones in terms of propagating importance along the graph.
- verdverm 4y agoOr just use PageRank as originally designed? Why make it more complex for a first try?