2000 character limit reached
Fighting Offensive Language on Social Media with Unsupervised Text Style Transfer (1805.07685v1)
Published 20 May 2018 in cs.CL and cs.LG
Abstract: We introduce a new approach to tackle the problem of offensive language in online social media. Our approach uses unsupervised text style transfer to translate offensive sentences into non-offensive ones. We propose a new method for training encoder-decoders using non-parallel data that combines a collaborative classifier, attention and the cycle consistency loss. Experimental results on data from Twitter and Reddit show that our method outperforms a state-of-the-art text style transfer system in two out of three quantitative metrics and produces reliable non-offensive transferred sentences.
- Cicero Nogueira dos Santos (31 papers)
- Igor Melnyk (28 papers)
- Inkit Padhi (31 papers)