M2A: A model-agnostic and metadata-free adversarial framework for unsupervised opinion summarization

Yanyue Zhang, Deyu Zhou, Zhenglin Wang, Yilong Lai

Abstract: Highlights•This marks the debut of adversarial framework in unsupervised opinion summarization.•Retrain the NLI module via unsupervised contrastive learning without annotations.•M2<math><msup is="true"><mrow is="true"></mrow><mrow is="true"><mn is="true">2</mn></mrow></msup></math>A outperforms some of SOTA in terms of faithfulness of summarization.