Task Details
We use the IMDB dataset and JOB benchmark The original JOB benchmark is included in the repository of the reference implementation. We change the predicates in the original JOB benchmark for each template to generate our evaluation set.
| Name | Description | Number of queries |
|---|---|---|
| JOB | The original JOB benchmark | 113 |
| JOB-sample | The sampled JOB benchmark | 983 |