Retrieving Relevant Context to Align Representations for Cross-lingual Event Detection

We study the problem of cross-lingual transfer learning for event detection (ED) where mod- els trained on a source language are expected to perform well on data for a new target lan- guage. Among a few recent works for this problem, the main approaches involve repre- sentation matching (e.g., adversarial training) that aims to eliminate language-specific fea- tures from the representations to achieve the language-invariant representations. However, due to the mix of language-specific features with event-discriminative context, representa- tion matching methods might also remove im- portant features for event prediction, thus hin- dering the performance for ED. To address this issue, we introduce a novel approach for cross-lingual ED where representations are aug- mented with additional context (i.e., not elim- inating) to bridge the gap between languages while enriching the contextual information to facilitate ED. At the core of our method in- volves a retrieval model that retrieves relevant sentences in the target language for an input sentence to compute augmentation representa- tions. Experiments on three languages demon- strate the state-of-the-art performance of our model for cross-lingual ED.