Could you create a PageUnrank algorithm for entities that have *lost* credibility with respect to other entities?
Serious question. It seems like it would be similar. The original PR treated a link as default positive valence on average…what if it’s not for key links?
Conversation
Replying to
Toy example. A graph with small world subgraphs P and Q. All intra-P or intra-Q links are positive valence. All P-Q cross-links are negative valence. Can you detect this condition? Can you detect outliers that have positive valence links from both P and Q?
1
7
Feels like just PR with weighted edges allowing for negative edges.
2
Replying to
The links tend to never be removed, so what could be used as a signal that would take precedence over these legacy links over time?
1
1
Show replies
Replying to
Something like this could adjust for regulatory capture, green washing, political reframing, woke word redefinition, and other subterfuge.
Or be a tool of it.
2


