This is one of the issues we're running into at my startup ApplyAll (automates job applications & the job search in general).
Our current best solution is to track when jobs get published/unpublished so that we can tell what's a repost (more likely to be a ghost job) vs a fresh job with high-intention to get the role filled.
I was talking to my co-founder this morning about collecting enough data so that we can analyze if it's even worth it for us to apply customers to re-posted jobs (there are legit reasons companies might do that) or if the hit rate is too low to bother (our kpi is interview-requests).