Embracing Domain Differences in Fake News: Cross-domain Fake News Detection using Multi-modal Data

Silva, Amila; Luo, Ling; Karunasekera, Shanika; Leckie, Christopher

doi:10.1609/aaai.v35i1.16134

Cited by 87 publications

(36 citation statements)

References 50 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Shu et al [25] develop a sentence-comment co-attention network to exploit news content and user comments to jointly capture check-worthy sentences and user comments for fake news detection. Silva et al [10] find that news records from different domains have significantly different word usage and propagation patterns. Therefore, the constructed model retains the knowledge of a specific domain to detect fake news from different domains effectively.…”

Section: Mixed Methodsmentioning

confidence: 99%

“…Although these methods are effective for fake news detection, they cannot be used alone to improve detection accuracy further. Therefore, some hybrid methods [10,11] come into being. By combining news content, propagation structure, social context, and source information for fake news detection, news features are fully characterized, greatly improving the detection effect.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Jointly multi-source information and local-global relations of heterogeneous network for rumor detection

Han

Zhao²,

Zhang³

et al. 2023

Front. Phys.

View full text Add to dashboard Cite

The widespread rumors on social media seriously disturb the social order, and we urgently need practical methods to detect rumors. Most existing deep learning methods focus on mining news text content, user information, and propagation features but ignore the rumor diffusion structural features. Rumors spread in a vertical chain and diffusion in a horizontal network. Both are essential features of rumors. In addition, existing models need more effective methods to extract higher-order features of multiple resource information. To address these problems, we propose a multi-source information heterogeneous graph model in this paper, called jointly Multi-Source information and Local-Global relationship of heterogeneous network model named MSLG. It extracts multi-source information such as rumors content, user information, propagation, and diffusion structure. Firstly, we extract the higher order semantic representation of rumors content by graph convolution network and integrate local relational attention to strengthen the critical semantic. At the same time, we construct the rumors and users as heterogeneous graphs to capture the propagation and diffusion structure of the rumors. We are finally fusing global relational attention to measure submodules’ importance. Experiments on two real-world datasets show that the proposed method achieves state-of-the-art results in fake news detection.

show abstract

Section: Mixed Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Jointly multi-source information and local-global relations of heterogeneous network for rumor detection

Han

Zhao²,

Zhang³

et al. 2023

Front. Phys.

View full text Add to dashboard Cite

show abstract

“…Closely related to our topic is fake news detection. [38] trains an event discriminator to overlook domain-specific knowledge under the multi-modal setting, [39] formulates domain-agnostic fake news detection as a continual learning problem, [37] studies the case with limited labeling budget, and [7,36] take advantage of auxiliary user descriptions and large-scale user corpus.…”

Section: Related Workmentioning

confidence: 99%

Probing Spurious Correlations in Popular Event-Based Rumor Detection Benchmarks

Wu¹,

Hooi²

2022

Preprint

View full text Add to dashboard Cite

As social media becomes a hotbed for the spread of misinformation, the crucial task of rumor detection has witnessed promising advances fostered by open-source benchmark datasets. Despite being widely used, we find that these datasets suffer from spurious correlations, which are ignored by existing studies and lead to severe overestimation of existing rumor detection performance. The spurious correlations stem from three causes: (1) event-based data collection and labeling schemes assign the same veracity label to multiple highly similar posts from the same underlying event;(2) merging multiple data sources spuriously relates source identities to veracity labels; and (3) labeling bias. In this paper, we closely investigate three of the most popular rumor detection benchmark datasets (i.e., Twitter15, Twitter16 and PHEME), and propose event-separated rumor detection to eliminate spurious cues. Under the event-separated setting, we observe that the accuracy of existing state-of-the-art models drops significantly by over 40%, becoming only comparable to a simple neural classifier. To better address this task, we propose Publisher Style Aggregation (PSA), a generalizable approach that aggregates publisher posting records to learn writing style and veracity stance. Extensive experiments demonstrate that our method outperforms existing baselines in terms of effectiveness, efficiency and generalizability.

show abstract

“…Currently, there are two main types of detection models from the perspective of topic distribution: in-topic models and cross-topic models (Ren et al, 2021). (1) In-topic models focus on scenarios where topics of the training and test data are same (Silva et al, 2021;Song et al, 2021), that is, the topics in the test set are seen. (2) Cross-topic models focus on scenarios where topics of the training and test data are different (Wang et al, 2018;Ren et al, 2021), which means the topics in the test data are unseen.…”

Section: Introductionmentioning

confidence: 99%

“…We argue that the reasons for the above phenomenon are as follows. (1) The performance of the in-topic model benefits from prior knowledge of the data (Ren et al, 2021;Silva et al, 2021), also known as topic-specific features, such as specific topic words. This prior knowledge cannot be transferred to new topics due to differences between topic features, resulting in poor generalisation of in-topic models to unseen topics.…”

Section: Introductionmentioning

confidence: 99%

Towards Robust False Information Detection on Social Networks with Contrastive Learning

et al. 2022

Proceedings of the 31st ACM International Conference on Information &Amp; Knowledge Management

View full text Add to dashboard Cite

Current works about false information detection based on conversation graphs on social networks focus primarily on two research streams from the standpoint of topic distribution: intopic and cross-topic techniques, which assume that the data topic distribution is identical or cross, respectively. This signifies that all test data topics are seen or unseen by the model. However, these assumptions are too harsh for actual social networks that contain both seen and unseen topics simultaneously, hence restricting their practical application. In light of this, this paper develops a novel open-topic scenario that is better suited to actual social networks. In this open-topic scenario, we empirically find that the existing models suffer from impairment in the detection performance for seen or unseen topic data, resulting in poor overall model performance. To address this issue, we propose a novel Contrastive Adversarial Learning Network, CALN, that employs an unsupervised topic clustering method to capture topic-specific features to enhance the model's performance for seen topics and an unsupervised adversarial learning method to align data representation distributions to enhance the model's generalisation to unseen topics. Experiments on two benchmark datasets and a variety of graph neural networks demonstrate the effectiveness of our approach.

show abstract

Embracing Domain Differences in Fake News: Cross-domain Fake News Detection using Multi-modal Data

Cited by 87 publications

References 50 publications

Jointly multi-source information and local-global relations of heterogeneous network for rumor detection

Jointly multi-source information and local-global relations of heterogeneous network for rumor detection

Probing Spurious Correlations in Popular Event-Based Rumor Detection Benchmarks

Towards Robust False Information Detection on Social Networks with Contrastive Learning

Contact Info

Product

Resources

About