Halo: Learning Semantics-Aware Representations for Cross-Lingual Information Extraction
This addresses cross-lingual information extraction for low-resource languages, but appears incremental as it builds on existing neural methods with a novel training technique.
The paper tackles the challenge of cross-lingual information extraction in low-resource scenarios by proposing Halo, a training method that enforces hidden states to generate tokens with consistent semantic tags, resulting in improved generalization without extra parameters.
Cross-lingual information extraction (CLIE) is an important and challenging task, especially in low resource scenarios. To tackle this challenge, we propose a training method, called Halo, which enforces the local region of each hidden state of a neural model to only generate target tokens with the same semantic structure tag. This simple but powerful technique enables a neural model to learn semantics-aware representations that are robust to noise, without introducing any extra parameter, thus yielding better generalization in both high and low resource settings.