Learning to Selectively Learn for Weakly-supervised Paraphrase Generation
This addresses the problem of generating high-quality paraphrases without large labeled datasets for NLP practitioners, though it is incremental as it builds on existing weak supervision and pre-trained models.
The paper tackles weakly-supervised paraphrase generation by using retrieval-based pseudo paraphrase expansion and a meta-learning framework to select valuable samples for fine-tuning BART, achieving significant improvements over unsupervised methods and performance comparable to supervised state-of-the-art approaches.
Paraphrase generation is a longstanding NLP task that has diverse applications for downstream NLP tasks. However, the effectiveness of existing efforts predominantly relies on large amounts of golden labeled data. Though unsupervised endeavors have been proposed to address this issue, they may fail to generate meaningful paraphrases due to the lack of supervision signals. In this work, we go beyond the existing paradigms and propose a novel approach to generate high-quality paraphrases with weak supervision data. Specifically, we tackle the weakly-supervised paraphrase generation problem by: (1) obtaining abundant weakly-labeled parallel sentences via retrieval-based pseudo paraphrase expansion; and (2) developing a meta-learning framework to progressively select valuable samples for fine-tuning a pre-trained language model, i.e., BART, on the sentential paraphrasing task. We demonstrate that our approach achieves significant improvements over existing unsupervised approaches, and is even comparable in performance with supervised state-of-the-arts.