More Context, Larger Models, or Moral Knowledge? A Systematic Study of Schwartz Value Detection in Political Texts
A new study published on arXiv investigates the effectiveness of context, model size, and moral knowledge for detecting Schwartz values in political texts. Researchers found that while increased context improved supervised DeBERTa encoders, it did not consistently benefit larger zero-shot LLMs. Retrieved moral knowledge proved more consistently useful across various models and context conditions, particularly for complex or socially situated values. The study suggests that optimal performance requires a joint evaluation of context, knowledge, and model family, rather than assuming larger models or longer inputs are universally superior. AI
IMPACT This research highlights nuanced factors beyond model size for effective text analysis, suggesting careful evaluation of context and knowledge integration for specialized NLP tasks.