I’m not the OP, and haven’t read much Marcus, but an analysis linking “chains of claims” together could be interesting, devaluing specific true claims when they are used to support a false claim. The claims dataset appears to evaluate each claim independently.