More Context, Larger Models, or Moral Knowledge? A Systematic Study of Schwartz Value Detection in Political Texts

TL;DR

This study systematically evaluates how context, explicit moral knowledge, and model size affect Schwartz value detection in political texts, revealing nuanced impacts on different model types and value categories.

Contribution

It provides a comprehensive comparison of context and knowledge integration methods across various models, highlighting that larger models and more context do not always lead to better performance.

Findings

Full-document context improves supervised models by 3.8-4.8 macro-F1 points.

Retrieved moral knowledge consistently benefits models, especially with early fusion.

Scaling models does not guarantee performance gains, and simple early fusion often outperforms complex methods.

Abstract

Detecting Schwartz values in political text is difficult because implicit cues often depend on surrounding arguments and fine-grained distinctions between neighboring values. We study when context and explicit moral knowledge help sentence-level value detection. Using the ValuesML/Touch{\'e} ValueEval format, we compare sentence, window, and full-document inputs; no-RAG and retrieval-augmented settings with a curated moral knowledge base; supervised DeBERTa-v3-base/large encoders; and zero-shot LLMs from 12B to 123B parameters. The results show that more context is not uniformly better: full-document context improves supervised DeBERTa encoders by 3.8--4.8 macro-F1 points over sentence-only input, but does not consistently help zero-shot LLMs. Retrieved moral knowledge is more consistently useful in matched comparisons, improving each tested model family and context condition under early fusion. However, scaling from DeBERTa-v3-base to large and from 12B to larger LLMs does not guarantee gains, and simple early fusion outperforms the tested late-fusion and cross-attention RAG variants for encoders. Per-value analyses show that context and retrieval help most for socially situated or conceptually confusable values. These findings suggest that value-sensitive NLP should evaluate context, knowledge, and model family jointly rather than treating longer inputs or larger models as universal improvements.