Can large language models extract operative standards from narrative operative reports in rectal cancer? — SciRadar