DeMis: Data-Efficient Misinformation Detection Using Reinforcement Learning
DeMis: Data-Efficient Misinformation Detection Using Reinforcement Learning
复制标题
DOI：
10.1007/978-3-031-26390-3_14
复制
发表时间：
2022
期刊：
影响因子：
--
通讯作者：
Kornraphop Kawintiranon;Lisa Singh
中科院分区：
文献类型：
--
作者：
Kornraphop Kawintiranon;Lisa Singh
深度学习方法对于许多自然语言处理任务来说是最先进的，包括错误信息检测。为了有效训练深度学习算法，大量的训练数据是必不可少的。不幸的是，虽然未标记的数据丰富，但人工标记的数据缺乏用于错误信息检测的数据。在本文中，我们提出了DeMis，一种新的强化学习（RL）框架，用于在资源受限的环境中检测Twitter上的错误信息，即有限的标记数据。主要的新颖之处来自(1)使用强化学习来识别高质量的弱标签，并与手动标记的数据一起使用，以联合训练分类器，以及(2)使用事实检查声明从未标记的推文中构建弱标签。我们通过经验证明了这种方法在当前技术状态下的优势，并证明了它在低资源环境下的有效性，比其他模型高出8% （F1分数）。我们还发现我们的方法对严重不平衡的数据具有更强的鲁棒性。最后，我们发布一个包含代码、训练模型和标记数据集的包。
Deep learning approaches are state-of-the-art for many natural language processing tasks, including misinformation detection. To train deep learning algorithms effectively, a large amount of training data is essential. Unfortunately, while unlabeled data are abundant, manually-labeled data are lacking for misinformation detection. In this paper, we propose DeMis, a novel reinforcement learning (RL) framework to detect misinformation on Twitter in a resource-constrained environment, i.e. limited labeled data. The main novelties result from (1) using reinforcement learning to identify high-quality weak labels to use with manually-labeled data to jointly train a classifier, and (2) using fact-checked claims to construct weak labels from unlabeled tweets. We empirically show the strength of this approach over the current state of the art and demonstrate its effectiveness in a low-resourced environment, outperforming other models by up to 8% (F1 score). We also find that our method is more robust to heavily imbalanced data. Finally, we publish a package containing code, trained models, and labeled data sets.