Language Model Priming for Cross-Lingual Event Extraction

We present a novel, language-agnostic approach to "priming" language models for the task of event extraction, providing particularly effective performance in low-resource and zero-shot cross-lingual settings. With priming, we augment input transformer stack's model differently depending on question(s) being asked at runtime. For instance, if is identify arguments trigger "protested", will provide that as part model, allowing it produce different representations candidate than when about "arrest" elsewhere same sentence. show by enabling better compensate deficits sparse noisy training data, our improves both argument detection classification significantly over state art setting.