一作ACL 2026Findings· Findings of ACL
Enhancing Factuality through Consensus and Consistency in Summarization Using Minimum B ayes Risk Decoding
Riza Setiawan Soetedjo · Yusuke Sakai · Hidetaka Kamigaito · Jingun Kwon · Manabu Okumura · Taro Watanabe
摘要
Improving the quality of model-generated summaries, especially factuality, the accuracy of a summary with respect to its source content, remains a challenge. While reranking could select the optimal output from multiple generated candidates, it is limited to only using the source as guidance, res...