Enhancing Factuality through Consensus and Consistency in Summarization Using Minimum Bayes Risk Decoding
Riza Setiawan Soetedjo, Yusuke Sakai, Hidetaka Kamigaito, Jingun Kwon, Manabu Okumura, Taro Watanabe · 2026
Read Enhancing Factuality through Consensus and Consistency in Summarization Using Minimum Bayes Risk Decoding by Riza Setiawan Soetedjo, Yusuke Sakai, Hidetaka Kamigaito, Jingun Kwon, Manabu Okumura, Taro Watanabe (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-29336.txt.
Description
Riza Soetedjo proposes ConSUM to improve the factuality of model summaries. Re-ranking could pick the best output from multiple candidates, but using only the source as guidance yields weak output. ConSUM reranks candidates by source consistency and shared support. Consensus comes from Minimum Bayes Risk decoding, while a fact-aware check compares each summary against the source. Testing shows it is on par with existing methods, and human tests prefer its outputs over other systems.
Catalog record
| CREATOR | Riza Setiawan Soetedjo, Yusuke Sakai, Hidetaka Kamigaito, Jingun Kwon, Manabu Okumura, Taro Watanabe |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 104988 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.29336