Rashomon Memory: Towards Argumentation-Driven Retrieval for Multi-Perspective Agent Memory
Albert Sadowski, Jarosław A. Chudziak · 2026
Read Rashomon Memory: Towards Argumentation-Driven Retrieval for Multi-Perspective Agent Memory by Albert Sadowski, Jarosław A. Chudziak (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-2604-03588.txt.
Description
Sadowski and Chudziak propose Rashomon Memory, a design where side-by-side goal-tuned agents encode events by their own goals and argue at query time. Each view keeps its own world model and knowledge graph. At lookup, views propose readings, critique each other with one-sided domain knowledge, and Dung's rules decide which proposals survive. The attack graph itself explains the outcome: it records the chosen reading, the other options weighed, and the grounds for turndown.
Catalog record
| CREATOR | Albert Sadowski, Jarosław A. Chudziak |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 41469 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.03588