Beyond Independent Labels: Schwartz-Geometry Decoding for Human Value Detection
2026-07-06 • Computation and Language
Computation and LanguageArtificial IntelligenceComputers and SocietyMachine Learning
AI summaryⓘ
The authors studied how to improve detecting human values in text by considering how values relate to each other, based on Schwartz's theory that values form a circle where nearby values are similar and opposite ones conflict. They tried adding this idea either during model training or after prediction. They found that adjusting predictions after training made the detected values fit the circular theory better without hurting accuracy, but changing training didn’t help much. This approach works specifically when using the real value sequence from Schwartz's theory and can help models respect the structure of value labels more naturally.
Schwartz valuesmulti-label classificationlabel space geometryDeBERTa-v3energy decoderstructured predictionmacro-F1micro-F1motivational continuumvalue detection
Authors
Víctor Yeste, Paolo Rosso
Abstract
Human value detection is commonly formulated as sentence-level multi-label classification over the 19 refined Schwartz values, typically predicted as independent labels. Schwartz theory, however, describes them as a circular motivational continuum, in which adjacent values are compatible and opposing values are in tension. We ask whether this structure can be operationalized as an explicit output-space geometry and used as a soft bias rather than a hard constraint. On a DeBERTa-v3-base classifier, we compare two ways of injecting it: training-time geometry-aware objectives and a post-hoc Schwartz-aware energy decoder that scores whole label sets jointly. Across five seeds, training-time geometry gives only limited gains-no larger for the true continuum than for a random ordering-whereas the decoder makes label sets more coherent with the continuum-on theory-aware coherence metrics we introduce-at no cost to Macro-F1 or Micro-F1 (held fixed by its selection rule). The gain is specific to the true Schwartz ordering: it does not appear for a random permutation or an empirical co-occurrence graph through the identical decoder. A bounded Qwen2.5-72B-Instruct diagnostic shows that supplying the continuum at inference shifts behavior but does not match supervised structured prediction. Theory-aware decoding thus offers a lightweight, controllable way to make value detection faithful to its label space.