PolERo: Studying Political Evasion in Romanian

2026-09-02

AI Summary

研究罗马尼亚政治回避问题，通过构建PolERo数据集并采用多种分类方法（包括微调编码器和少样本提示）来解决跨语言迁移和模型分类挑战。

Abstract

Political evasion refers to responses that engage with a question while withholding the requested information. Recent NLP work frames political evasion as a classification task using a two-level taxonomy of response clarity and fine-grained evasion strategies. Existing work on response clarity and evasion classification is limited to English, leaving open whether the taxonomy and model behavior transfer across languages and political contexts. We introduce PolERo, a dataset of 3,574 human-annotated question-answer pairs extracted from official transcripts of five Romanian presidents. We evaluate multiple classification approaches on both datasets under matched conditions, including TF-IDF baselines, fine-tuned encoder models, a proposed sliding-window encoder, and zero/few-shot LLM prompting. We study cross-lingual transfer through joint bilingual training and machine-translation-based data augmentation. Our results indicate that fine-tuned encoders are competitive, cross-lingual transfer is asymmetric, and ambivalent evasion categories involving pragmatic cues remain the main challenge across all model families.

Problem

• political evasion
• response clarity
• cross-lingual transfer
• Romanian

Innovation

• cross-lingual transfer
• sliding-window encoder
• few-shot LLM prompting