After the clustering, it then identifies points of agreement across clusters and surfaces those for voting to verify. That’s in the article on the linked site:

After the clustering, it then identifies points of agreement across clusters and surfaces those for voting to verify. That’s in the article on the linked site: “The algorithm then identifies “consensus statements" about which all the diverse clusters agree.” There’s also some notes about “group-informed consensus” on the algorithms page that might be interesting for you: https://compdemocracy.org/group-informed-consensus/ https://compdemocracy.org/group-informed-consensus/ And the “comment routing” algorithm: https://compdemocracy.org/comment-routing/ https://compdemocracy.org/comment-routing/