Unsupervised Adversarial Domain Adaptation for Implicit Discourse Relation Classification
This work addresses a domain-specific challenge in natural language processing for researchers and practitioners dealing with limited annotated data in discourse analysis.
The paper tackled the problem of classifying implicit discourse relations when training data is scarce by using unsupervised adversarial domain adaptation from explicit relations, achieving performance that outperforms prior works and other adversarial benchmarks.
Implicit discourse relations are not only more challenging to classify, but also to annotate, than their explicit counterparts. We tackle situations where training data for implicit relations are lacking, and exploit domain adaptation from explicit relations (Ji et al., 2015). We present an unsupervised adversarial domain adaptive network equipped with a reconstruction component. Our system outperforms prior works and other adversarial benchmarks for unsupervised domain adaptation. Additionally, we extend our system to take advantage of labeled data if some are available.