Great idea! The challenges one can imagine with that are: 1) human reviews are expensive 2) the citation might not have anything to do with the reason why the

Great idea! The challenges one can imagine with that are: 1) human reviews are expensive 2) the citation might not have anything to do with the reason why the cited paper was retracted. Now, if it all could be automated (say with some emerging technology that can grok the context of the citation and say classify it into flag/no-flag buckets say), then we are cooking with some real fire. :-)