语码转换模式可作为改进下游 NLP 应用性能的有效途径:以幽默、讽刺与仇恨言论检测为例
计算与语言
2020-05-06 v1
摘要
在本文中,我们展示了如何利用语码转换模式来改进各种下游 NLP(自然语言处理)应用。具体而言,我们编码不同的转换特征以改进幽默、讽刺和仇恨言论检测任务。我们认为这一简单的语言学观察也有潜力有助于改进其他类似的 NLP 应用。
引用
@article{arxiv.2005.02295,
title = {Code-switching patterns can be an effective route to improve performance of downstream NLP applications: A case study of humour, sarcasm and hate speech detection},
author = {Srijan Bansal and Vishal Garimella and Ayush Suhane and Jasabanta Patro and Animesh Mukherjee},
journal= {arXiv preprint arXiv:2005.02295},
year = {2020}
}
备注
This work is accepted as a short paper in the proceedings of ACL 2020