Comment by craydandy

I've found the same. It sounds great and applicable but could not find any practical application with real data.

Even when applied to large number of real companies' journal entries, the amount of false positives is overwhelming.