EANN: Event Adversarial Neural Networks for Multi-Modal Fake News Detection

By Yaqing Wang, Fenglong Ma, Zhiwei Jin, Ye Yuan, Guangxu Xun, Kishlay Jha, Lu Su, Jing Gao. Knowledge Discovery and Data Mining, 2018

1,234 citations recorded by the source corpus.

As news reading on social media becomes more and more popular, fake news becomes a major issue concerning the public and government. The fake news can take advantage of multimedia content to mislead readers and get dissemination, which can cause negative effects or even manipulate the public events. One of the unique challenges for fake news detection on social media is how to identify fake news on newly emerged events. Unfortunately, most of the existing approaches can hardly handle this challenge, since they tend to learn event-specific features that can not be transferred to unseen events. In order to address this issue, we propose an end-to-end framework named Event Adversarial Neural Network (EANN), which can derive event-invariant features and thus benefit the detection of fake news on newly arrived events. It consists of three main components: the multi-modal feature extractor, the fake news detector, and the event discriminator. The multi-modal feature extractor is responsible for extracting the textual and visual features from posts. It cooperates with the fake news detector to learn the discriminable representation for the detection of fake news. The role of event discriminator is to remove the event-specific features and keep shared features among events. Extensive experiments are conducted on multimedia datasets collected from Weibo and Twitter. The experimental results show our proposed EANN model can outperform the state-of-the-art methods, and learn transferable feature representations.

Subfields: Artificial Intelligence, Information Systems, Sociology and Political Science

View Paper: https://www.semanticscholar.org/paper/5c107e9d3e09bc33c416cd214d3760ba8c32a470

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Xinyi Zhou
2. Kai Shu
3. Huan Liu
4. Philip S. Yu
5. Lichao Sun
6. Suhang Wang
7. Reza Zafarani
8. Yingtong Dou

In Information Systems

1. Xinyi Zhou
2. Kai Shu
3. Huan Liu
4. Reza Zafarani
5. Xinyi Zhou
6. Preslav Nakov
7. Suhang Wang
8. Min-Yen Kan

In Sociology and Political Science

1. Xinyi Zhou
2. Kai Shu
3. Reza Zafarani
4. Xinyi Zhou
5. Huan Liu
6. Yi-Ju Lu
7. Cheng-Te Li
8. Preslav Nakov

Citations in the corpus

• Beyond News Contents: The Role of Social Context for Fake News Detection (2017, background): “Recently, deep learning models are applied to learn the temporal and linguistic representation of news [11, 30, 35].”
• Fake News: A Survey of Research, Detection Methods, and Opportunities (2018, methodology): “…across topics, domains, language, and the like, as well as the evolution of deceptive writing style, where Generative Adversarial Networks (GANs) [Wang et al. 2018] and Recurrent Neural Networks (RNNs) [Chen et al. 2017] have played or can play to their strengths; (3) verification efficiency,…”
• Cross-modal Ambiguity Learning for Multimodal Fake News Detection (2022, methodology): “This work focuses on fake news detection with text and image, so we remove social information from multimodal baselines for a fair comparison; M2: EANN [31], which consists of two related tasks: event discrimination and fake news detection.”
• A Survey of Fake News (2020, methodology): “…related to dynamic (real-time) KG construction should be developed to facilitate timely updates of the ground truth; (2) feature compatibility [Wang et al. 2018], specifically, features that can capture the commonalities of deceptive writing style across topics, domains, languages, and the…”
• Multimodal Fusion with Co-Attention Networks for Fake News Detection (2021, methodology): “In EANN, event iden-tification is an auxiliary task, and event labels are not in original datasets.”
• Mining Dual Emotion for Fake News Detection (2019, background): “We suppose that the duplication may increase the risk of data leakage when splitting training and testing sets and make models tend to learn some event-specific features[42] (as they may repeat multiple times in the training process), which limits the generalizability of models.”
• dEFEND: Explainable Fake News Detection (2019, background): “Recently, research also focuses on challenging problems of fake news detection, such as fake news early detection by adversarial learning [45] and user response generating [35], semi-supervised detection [11] and unsupervised detection [15, 49], and explainable detection of fake news through meta…”
• SpotFake: A Multi-modal Framework for Fake News Detection (2019, methodology): “Performance gain on Weibo dataset over EANN- and EANN is 9.73% and 6.53% respectively.”
• GCAN: Graph-aware Co-Attention Networks for Explainable Fake News Detection on Social Media (2020, background): “Wang et al. (2018) develop an event adversarial neural network to learn transferable features by removing the event-specific features, along with convolutional neural networks to extract textual and visual features. dEFEND (Shu et al., 2019a) jointly learns the sequential effect of response comments…”
• Hierarchical Multi-modal Contextual Attention Network for Fake News Detection (2021, background): “For fairness of the comparison, we conduct experiments with a simplified version of EANN that excludes the post discriminator.”
• FANG: Leveraging Social Context for Fake News Detection Using Graph Representation (2020, methodology): “Textual [6, 34, 39, 44] and visual [18, 43] features have been widely used to model news article contents, either by feature extraction, unsupervised semantics encoding, or learned representation.”
• User Preference-aware Fake News Detection (2021, background): “In addition, computational approaches using feature engineering or deep learning have shown many promising results [3, 12, 23, 31].”