Detecting Hate Speech in Cross-Lingual and Multi-lingual Settings Using Language Agnostic Representations

Rodríguez, Sebastián E.; Allende‐Cid, Héctor; Allende, Héctor

doi:10.1007/978-3-030-93420-0_8

Cited by 4 publications

(2 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We focus on two aspects: 1) considering languages that are typologically distinct from English and 2) with little effort. Previous works focus on languages that are similar to English, such as European languages (Stappen et al, 2020;Nozza, 2021;Rodríguez et al, 2021;Firmino et al, 2021;Zia et al, 2022;Castillo-López et al, 2023). In contrast, we analyze languages that are more different from English.…”

Section: Introductionmentioning

confidence: 99%

Vicinal Risk Minimization for Few-Shot Cross-lingual Transfer in Abusive Language Detection

De la Peña Sarracén,

Rosso,

Litschko

et al. 2023

Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing

View full text Add to dashboard Cite

Cross-lingual transfer learning from highresource to medium and low-resource languages has shown encouraging results. However, the scarcity of resources in target languages remains a challenge. In this work, we resort to data augmentation and continual pre-training for domain adaptation to improve cross-lingual abusive language detection. For data augmentation, we analyze two existing techniques based on vicinal risk minimization and propose MIXAG, a novel data augmentation method which interpolates pairs of instances based on the angle of their representations. Our experiments involve seven languages typologically distinct from English and three different domains. The results reveal that the data augmentation strategies can enhance fewshot cross-lingual abusive language detection. Specifically, we observe that consistently in all target languages, MIXAG improves significantly in multidomain and multilingual environments. Finally, we show through an error analysis how the domain adaptation can favour the class of abusive texts (reducing false negatives), but at the same time, declines the precision of the abusive language detection model.

show abstract

Section: Introductionmentioning

confidence: 99%

Vicinal Risk Minimization for Few-Shot Cross-lingual Transfer in Abusive Language Detection

De la Peña Sarracén,

Rosso,

Litschko

et al. 2023

Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing

View full text Add to dashboard Cite

show abstract

“…We focus on two aspects: 1) considering languages that are typologically distinct from English and 2) with little effort. Previous works focus on languages that are similar to English, such as European languages (Stappen et al, 2020b;Nozza, 2021;Rodríguez et al, 2021;Firmino et al, 2021;Zia et al, 2022;Castillo-López et al, 2023). In contrast, we analyze languages that are more different from English.…”

Section: Introductionmentioning

confidence: 99%

On the Keyword Extraction and Bias Analysis, Graph-based Exploration and Data Augmentation for Abusive Language Detection in Low-Resource Settings

Peña Sarracén

View full text Add to dashboard Cite

Many thanks go to my colleagues and fellow researchers at the Universitat Politècnica de València, whose collaboration and support have been both enjoyable and intellectually enriching. The synergy between academia and industry fostered by Symanto has been instrumental in achieving the objectives of this thesis.I extend my appreciation to the reviewers of this thesis Dr. Arkaitz Zubiaga (Queen Mary University of London), Prof. Rafael Valencia (Universidad de Murcia), and Dr. Tommaso Caselli (Rijksuniveristeit Groningen); thank you very much for your work on my thesis. I would also like to thank the members of the PhD committee of this thesis Dr. Arkaitz Zubiaga (Queen Mary University of London), Dra. Sara Tonelli (Fondazione Bruno Kessler), and Dra. Aitziber Atutxa (University of the Basque Country).Special thanks go to my family for their tireless encouragement and understanding. Their love and support have been my pillars throughout this challenging yet rewarding endeavor. Also to the friends who have taken an interest in my emotional and physical well-being during my doctoral research.Last but not least, I would like to thank all the participants and individuals who generously contributed their time. I am truly grateful.

show abstract

Forecasting National-Level Self-Harm Trends With Social Networks

et al. 2023

View full text Add to dashboard Cite

Detecting Hate Speech in Cross-Lingual and Multi-lingual Settings Using Language Agnostic Representations

Cited by 4 publications

References 11 publications

Vicinal Risk Minimization for Few-Shot Cross-lingual Transfer in Abusive Language Detection

Vicinal Risk Minimization for Few-Shot Cross-lingual Transfer in Abusive Language Detection

On the Keyword Extraction and Bias Analysis, Graph-based Exploration and Data Augmentation for Abusive Language Detection in Low-Resource Settings

Forecasting National-Level Self-Harm Trends With Social Networks

Contact Info

Product

Resources

About