Formality style transfer 를 하고자 하며, parallel data가 부족하다는 것에 문제점를 해결하고자 합니다.
formality를 분류할 수 있는 모델을 이용해서 데이터를 만들고, 소량의 parallel data 를 같이 사용하였습니다.
이 방식을 통해서 Formality 테스크에서 SOTA 를 달성하였습니다.
저자는 이 방식이 다른 unsupervised style transfer 에도 사용될 수 있다는 점을 보였습니다. (3개의 benchmarks 에서도 SOTA를 수정했음)
Abstract (요약) 🕵🏻♂️
Formality style transformation is the task of modifying the formality of a given sentence without changing its content. Its challenge is the lack of large-scale sentence-aligned parallel data. In this paper, we propose an omnivorous model that takes parallel data and formality-classified data jointly to alleviate the data sparsity issue. We empirically demonstrate the effectiveness of our approach by achieving the state-of-art performance on a recently proposed benchmark dataset of formality transfer. Furthermore, our model can be readily adapted to other unsupervised text style transfer tasks like unsupervised sentiment transfer and achieve competitive results on three widely recognized benchmarks.
이 논문을 읽어서 무엇을 배울 수 있는지 알려주세요! 🤔
classification 모델을 학습해서 이를 style transfer 용 paraller corpus 를 만드는데 사용하는 방식에 대한 인사이트를 얻을 수 있습니다.
어떤 내용의 논문인가요? 👋
Abstract (요약) 🕵🏻♂️
Formality style transformation is the task of modifying the formality of a given sentence without changing its content. Its challenge is the lack of large-scale sentence-aligned parallel data. In this paper, we propose an omnivorous model that takes parallel data and formality-classified data jointly to alleviate the data sparsity issue. We empirically demonstrate the effectiveness of our approach by achieving the state-of-art performance on a recently proposed benchmark dataset of formality transfer. Furthermore, our model can be readily adapted to other unsupervised text style transfer tasks like unsupervised sentiment transfer and achieve competitive results on three widely recognized benchmarks.
이 논문을 읽어서 무엇을 배울 수 있는지 알려주세요! 🤔
레퍼런스의 URL을 알려주세요! 🔗
https://arxiv.org/abs/1903.06353