An Annotation Scheme and Classifier for Personal Facts in Dialogue
Konstantin Zaitsev · 2026
Read An Annotation Scheme and Classifier for Personal Facts in Dialogue by Konstantin Zaitsev (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-2605-10339.txt.
Description
Konstantin Zaitsev improves how chat systems recall personal facts about users. Chat systems pile up facts such as tastes and ties, but past labeling schemes, notably PeaCoK, limit tidy storage and quality filtering. His 2026 paper offers a wider labeling scheme for personal fact sorting, with new groups and traits for carrying a chat forward.
The scheme adds Demographics and Possessions as groups, plus Duration, Validity, and Followup as traits. That enables tidy storage, quality filtering, and spotting facts fit to continue a chat. Zaitsev hand-labeled 2,779 facts from Multi-Session Chat and trained a multi-head sorter on transformer encoders. With the Gemma-300M encoder it hit 81.6 percent macro F1, beating the best few-shot language-model baseline, at 72.92 percent, by nearly 9 points while using far less computing power. Data and sorter are public.
Catalog record
| CREATOR | Konstantin Zaitsev |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 70 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.10339