Suggestions for backend

[Phippsy]

• Reputation metrics: maintain a 'blacklist' of known debunked and/or validated articles to influence reputation?
• Contribute to publisher / sharer categorisations using tf-idf on entities to identify and quantify high-affinity topics. Use 'publisher' as the defined collection for tf-idf.
• Surface 'most reputable sources by entity' for any given user-selected entity, to help people find the most trustworthy source?

[MarkEdmondson1234]

Reputation metrics: maintain a 'blacklist' of known debunked and/or validated articles to influence reputation?

Yes, a liberal hippy may get very low reputation scores for news sources, but a right wing fascist could get very high scores for the same. Everyone will have a chance to input a whitelist and blacklist of articles they disagree with, and then through network associations with who share or link to those lists, build up a rank of new sources they come across. Hopefully also update those scores as they go.

Contribute to publisher / sharer categorisations using tf-idf on entities to identify and quantify high-affinity topics. Use 'publisher' as the defined collection for tf-idf.

This sounds similar to the entity score "salience" in the API responses got so far?

Surface 'most reputable sources by entity' for any given user-selected entity, to help people find the most trustworthy source?

Cool, yes.