TY  - RPRT
TI  - Optimising Factual Consistency in Summarisation via Preference Learning from Multiple Imperfect Metrics
AU  - Yuxuan Ye
AU  - Raul Santos-Rodriguez
AU  - Edwin Simpson
PY  - 2026
DO  - 10.18653/v1/2025.findings-emnlp.940
UR  - https://arxiv.org/abs/2605.26840
ID  - 2605.26840
ER  -