Do Schwartz Higher-Order Values Help Sentence-Level Human Value Detection? When Hard Gating Hurts Journal: Knowledge-Based Systems
Víctor Yeste, Paolo Rosso · 2026
Read Do Schwartz Higher-Order Values Help Sentence-Level Human Value Detection? When Hard Gating Hurts Journal: Knowledge-Based Systems by Víctor Yeste, Paolo Rosso (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-2602-00913.txt.
Description
Víctor Yeste and colleagues find the best gains come from calibration and ensembling, with threshold tuning lifting Social Focus versus Personal Focus by 0.16. They ask whether Schwartz higher-order categories help sentence-level human value detection. They compare direct trained transformers, hard higher-order chains, Presence chains, compact instruction LLMs, QLoRA, and cheap upgrades like threshold tuning and small teams. Higher-order categories are learnable: Growth versus Self-Protection is the easiest bipolar pair. Layered gating does not often improve the end task, and compact LLMs underperform trained encoders.
Catalog record
| CREATOR | Víctor Yeste, Paolo Rosso |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 64094 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602.00913