Enhancing the Detection of Misogynistic Content in Social Media by Transferring Knowledge From Song Phrases

Calderon-Suarez, Ricardo; Ortega-Mendoza, Rosa M.; Montes-y-Gómez, Manuel; Toxqui-Quitl, C.; Márquez-Vera, Marco Antonio

doi:10.1109/access.2023.3242965

Cited by 8 publications

(2 citation statements)

References 51 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Research has shown that pre-trained language models, such as the Transformer [16]-based bidirectional encoder BERT [17], can significantly improve the performance of natural language processing tasks with limited data. Calderón-Suárez R [18] introduced a novel data augmentation method that utilizes lyrics to improve the generalization ability of the method and enhance its performance. For long-tail entities and relations in knowledge graphs, traditional methods that utilize structural information often yield randomly initialized entity and relation representations, which leads to a sharp decline in reasoning performance on long-tail entities.…”

Section: Text-description-based Approachmentioning

confidence: 99%

Knowledge Graph Completion Using a Pre-Trained Language Model Based on Categorical Information and Multi-Layer Residual Attention

Rao,

Wang,

Guo

et al. 2024

Applied Sciences

View full text Add to dashboard Cite

Knowledge graph completion (KGC) utilizes known knowledge graph triples to infer and predict missing knowledge, making it one of the research hotspots in the field of knowledge graphs. There are still limitations in generating high-quality entity embeddings and fully understanding the contextual information of entities and relationships. To overcome these challenges, this paper introduces a novel pre-trained language model-based method for knowledge graph completion that significantly enhances the quality of entity embeddings by integrating entity categorical information with textual descriptions. Additionally, this method employs an innovative multi-layer residual attention network in combination with PLMs, deepening the understanding of the joint contextual information of entities and relationships. Experimental results on the FB15k-237 and WN18RR datasets demonstrate that our proposed model significantly outperforms existing baseline models in link prediction tasks.

show abstract

Section: Text-description-based Approachmentioning

confidence: 99%

Knowledge Graph Completion Using a Pre-Trained Language Model Based on Categorical Information and Multi-Layer Residual Attention

Rao,

Wang,

Guo

et al. 2024

Applied Sciences

View full text Add to dashboard Cite

show abstract

“…These studies reflect the expanding scope of content moderation to include geographical and linguistic analyses. Finally, Bacha et al [24] and Calderón-Suarez et al [25] contributed to offensive text detection in unstructured data for heterogeneous social media and enhancing the detection of misogynistic content in social media scenarios. Their research underscores the ongoing efforts to create safer and more inclusive online environments for different scenarios.…”

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

Enhancing misogyny detection in bilingual texts using explainable AI and multilingual fine-tuned transformers

Hashmi,

Yayilgan,

Yamin

et al. 2024

Complex Intell. Syst.

View full text Add to dashboard Cite

Gendered disinformation undermines women’s rights, democratic principles, and national security by worsening societal divisions through authoritarian regimes’ intentional weaponization of social media. Online misogyny represents a harmful societal issue, threatening to transform digital platforms into environments that are hostile and inhospitable to women. Despite the severity of this issue, efforts to persuade digital platforms to strengthen their protections against gendered disinformation are frequently ignored, highlighting the difficult task of countering online misogyny in the face of commercial interests. This growing concern underscores the need for effective measures to create safer online spaces, where respect and equality prevail, ensuring that women can participate fully and freely without the fear of harassment or discrimination. This study addresses the challenge of detecting misogynous content in bilingual (English and Italian) online communications. Utilizing FastText word embeddings and explainable artificial intelligence techniques, we introduce a model that enhances both the interpretability and accuracy in detecting misogynistic language. To conduct an in-depth analysis, we implemented a range of experiments encompassing classic machine learning methodologies and conventional deep learning approaches to the recent transformer-based models incorporating both language-specific and multilingual capabilities. This paper enhances the methodologies for detecting misogyny by incorporating incremental learning for cutting-edge datasets containing tweets and posts from different sources like Facebook, Twitter, and Reddit, with our proposed approach outperforming these datasets in metrics such as accuracy, F1-score, precision, and recall. This process involved refining hyperparameters, employing optimization techniques, and utilizing generative configurations. By implementing Local Interpretable Model-agnostic Explanations (LIME), we further elucidate the rationale behind the model’s predictions, enhancing understanding of its decision-making process.

show abstract

Enhancing the Detection of Misogynistic Content in Social Media by Transferring Knowledge From Song Phrases

Cited by 8 publications

References 51 publications

Knowledge Graph Completion Using a Pre-Trained Language Model Based on Categorical Information and Multi-Layer Residual Attention

Knowledge Graph Completion Using a Pre-Trained Language Model Based on Categorical Information and Multi-Layer Residual Attention

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

Enhancing misogyny detection in bilingual texts using explainable AI and multilingual fine-tuned transformers

Contact Info

Product

Resources

About