Things that we can do to improve the oracles dataset are:
Remove duplicated oracle datapoints which have different oracles (one has an oracle and the other has no oracle). I think Davide is already taking care of this, they should be very few cases.
Remove wrong oracles if no oracle can be generated for a particular oracle datapoint (i.e., replace the oracle with “;”).
Fix wrong oracles if an alternative oracle can be generated for a particular oracle datapoint.
Add new oracles for those oracle datapoints where no oracle was generated (i.e., it is “;”) but an oracle is possible.
Things that we can do to improve the oracles dataset are: