Thank you very much for sharing this dataset.
In the file discox.json (https://huggingface.co/datasets/ByteDance-S), there are prompt and original text for each record, but we cannot find any human translation.
In table 3 of the paper, there is a score for human expert translation. How did you obtain the human translation?