Towards Open Domain Event Trigger Identification using Adversarial Domain Adaptation

TL;DR

This paper introduces an adversarial domain adaptation framework for open domain event trigger identification, enabling models to generalize across domains without requiring labeled target data.

Contribution

It presents a novel adversarial training approach that creates domain-invariant representations for event trigger detection, improving out-of-domain performance.

Findings

ADA improves F1 scores by 3.9 on out-of-domain data

BERT-A achieves 44-49 F1 without labeled target data

Fine-tuning with 1% labeled data and self-training boosts F1 to 51.5 and 67.2

Abstract

We tackle the task of building supervised event trigger identification models which can generalize better across domains. Our work leverages the adversarial domain adaptation (ADA) framework to introduce domain-invariance. ADA uses adversarial training to construct representations that are predictive for trigger identification, but not predictive of the example's domain. It requires no labeled data from the target domain, making it completely unsupervised. Experiments with two domains (English literature and news) show that ADA leads to an average F1 score improvement of 3.9 on out-of-domain data. Our best performing model (BERT-A) reaches 44-49 F1 across both domains, using no labeled target data. Preliminary experiments reveal that finetuning on 1% labeled data, followed by self-training leads to substantial improvement, reaching 51.5 and 67.2 F1 on literature and news respectively.