Rumor Spreading Model Considering the Roles of Online Social Networks and Information Overload

Fu, Jinlong; Song, Yan; Feng, Yike

doi:10.1109/access.2023.3328396

Cited by 1 publication

(1 citation statement)

References 72 publications

(74 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Khan et al [41] focused on visual user-generated content verification in journalism, reflecting the growing importance of visual content in the realm of fake news. Fu et al [42] examined rumor spreading models considering the roles of online social networks and information overload, providing insights into the behavioral aspects of misinformation dissemination operations. Further extending the scope of research, Wang et al [43], Hu et al [44], and Wang et al [45] explored various aspects of multi-modal fake news detection, including the use of transformer networks and causal inference.…”

Section: Related Studymentioning

confidence: 99%

MSCMGTB: A Novel Approach for Multimodal Social Media Content Moderation Using Hybrid Graph Theory and Bio-Inspired Optimization

Arya,

Pandey,

Gopal Krishna Patro

et al. 2024

IEEE Access

View full text Add to dashboard Cite

In an era where social media platforms burgeon with diverse content, effective content moderation becomes imperative to filter harmful materials. Traditional methods often grapple with the dual challenges of accuracy and computational efficiency levels. These conventional approaches typically rely on either text-based or image-based analysis, neglecting the complex interplay of multimodal content prevalent in social media scenarios. This limitation leads to suboptimal content filtering, often missing contextually nuanced or visually deceptive harmful content sets. Addressing these challenges, the proposed work introduces an innovative, hybrid deep learning model that integrates Convolutional Neural Networks (CNNs) and Transformers for nuanced visual content extraction from social media posts. CNNs adeptly handle image features, while Transformers enrich the analysis with contextual textual understanding, thereby enhancing accuracy in identifying harmful contents. Additionally, a Bi-directional Attention Mechanism (BAM) is employed for efficient text-visual fusion, dynamically prioritizing relevant information and reducing computational loads. The fusion process is further refined using Genetic Algorithms (GAs) for hyperparameter optimization, streamlining the model's performance levels. The model's robustness is demonstrated through its ability to discern complex intra and inter-modal relationships within social media content, facilitated by Graph Neural Networks (GNNs). Moreover, the model's scalability and deployment efficiency on social media platforms are ensured through quantization and pruning techniques, balancing accuracy with computational practicality levels. Empirical testing on datasets from Google, Facebook, and Kaggle reveals the model's superior performance over existing methods, with notable improvements in precision, accuracy, recall, AUC, specificity, and response delay in detecting harmful contents. The model also excels in preempting potential harmful content posters, offering enhanced pre-emption metrics.

show abstract