Few-Shot, No Problem: Descriptive Continual Relation Extraction
Nguyen Xuan Thanh, Anh Duc Le, Quyen Tran, Thanh-Thien Le, Linh Ngo Van, Thien Huu Nguyen · 2025
Read Few-Shot, No Problem: Descriptive Continual Relation Extraction by Nguyen Xuan Thanh, Anh Duc Le, Quyen Tran, Thanh-Thien Le, Linh Ngo Van, Thien Huu Nguyen (2025) 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-2502-20596.txt.
Description
Nguyen Thanh and co-authors solve in 2025 few-shot continual relation extraction, where models learn new text relations from tiny samples over time. Old memory tools overfit small samples and forget old relations. Scarce data also block the usual tricks for boosting sample sets.
A large model first writes a plain description for each relation. A twin encoder then learns from both samples and descriptions to enrich their forms. At test time, each sample pulls its best relation by a fused rank score over descriptions and class cores. Tests across data sets beat past tools while holding old skill across task steps.
Catalog record
| CREATOR | Nguyen Xuan Thanh, Anh Duc Le, Quyen Tran, Thanh-Thien Le, Linh Ngo Van, Thien Huu Nguyen |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 34 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-20596