When developing a KBP system, we often predict provenances for hop0 entities that haven't been seen in the evaluation dataset. The CS-GenerateQueries script generates query ids that depend on the provenance and hence produce new query ids that are not comparable with those in the evaluation data.
Is it possible to handle this issue in the CS-Score script?
When developing a KBP system, we often predict provenances for hop0 entities that haven't been seen in the evaluation dataset. The
CS-GenerateQueriesscript generates query ids that depend on the provenance and hence produce new query ids that are not comparable with those in the evaluation data.Is it possible to handle this issue in the
CS-Scorescript?