DisSent: Sentence Representation Learning from Explicit Discourse Relations
This work addresses the need for cost-effective sentence representation learning in natural language understanding, offering a novel dataset curation method that reduces reliance on manual annotation.
The paper tackled the problem of learning effective sentence representations by curating a high-quality dataset using explicit discourse relations from dependency parsing and rule-based rubrics, achieving state-of-the-art results on the Penn Discourse Treebank's implicit relation prediction task and high performance on SentEval transfer tasks.
Learning effective representations of sentences is one of the core missions of natural language understanding. Existing models either train on a vast amount of text, or require costly, manually curated sentence relation datasets. We show that with dependency parsing and rule-based rubrics, we can curate a high quality sentence relation task by leveraging explicit discourse relations. We show that our curated dataset provides an excellent signal for learning vector representations of sentence meaning, representing relations that can only be determined when the meanings of two sentences are combined. We demonstrate that the automatically curated corpus allows a bidirectional LSTM sentence encoder to yield high quality sentence embeddings and can serve as a supervised fine-tuning dataset for larger models such as BERT. Our fixed sentence embeddings achieve high performance on a variety of transfer tasks, including SentEval, and we achieve state-of-the-art results on Penn Discourse Treebank's implicit relation prediction task.