Comunicación‎ > ‎

Facebook Research financia dos proyectos del ISISTAN sobre tratamiento de desinformación en medios sociales.

publicado a la‎(s)‎ 4 nov. 2019 11:29 por Administrador Web


La Dra. Antonela Tommasel y la Dra. Daniela Godoy obtuvieron proyectos en los Content Policy Research on Social Media Platforms y Online Safety Benchmark Research research awards. El primero de ellos en colaboración con el grupo del Dr. Arkaitz Zubiaga de la Queen Mary University of London.


Los proyectos financiados son los siguientes:


Hate Speech Is in the Eye of the Beholder: Exploring Bias on Hate Perception
Content Policy Research on Social Media Platforms
https://research.fb.com/blog/2019/09/announcing-the-winners-of-phase-two-content-policy-research-awards/

This project proposes a comprehensive study of bias in hate speech datasets resulting from the demographics of human annotators and its consequences in detection methods. Particularly, the study comprises three main phases: (1) assessing the bias of existing hate speech datasets through cross-validation, (2) capturing the different perspectives over utterances based on diverse demographic backgrounds and evaluating its impact on the hatefulness assessment, and (3) exploring the effect of perceived popularity in the assessment of hatefulness in combination with the demographic characteristics.

Faking It! A fake news multi-sourced dataset powered by Social Media
Online Safety Benchmark Research

Nowadays social media enriches the life and activities of its users, thus giving rise to new forms of communication and interaction. Nonetheless, at the same time social media also represents the ideal environment for undesirable phenomena, such as the dissemination of unwanted or unreliable content, and misinformation. Consequently, in the last few years, the research on misinformation has received increasing attention. Even though some computational solutions have been presented, the lack of a common ground and public datasets has become one of the major barriers. Not only datasets are rare, but also, they are mostly limited to the actual shared text, neglecting the importance of other features, such as social content and temporal information. In this scenario, this project proposes the creation of a publicly available dataset, comprising multi-sourced data and including diverse features related not only to the textual and multimedia content, but also to the social context of news and their temporal information. To allow studying the characteristics of fake news in comparison to those of real news, the dataset will include both real and fake news. It is expected that this dataset would not only allow tackling the task of fake news and accounts detection, but also studying their evolution and engagement cycle, which, in turn, can foster the development of mitigation and debunking techniques. As social media TOS forbids to share the actual content, we will develop a companion Java tool that will allow to download all data included in the dataset.


Comments