Pamplona, April 29 (EFE).- A new tool, “Xenometer”, already experimentally monitors xenophobic messages on social networks in Spain, an initiative that in the hands of the police or institutions can help stop them and contribute to the education of the population so that they do not proliferate.
Sociologist Sergio García-Magariño, coordinator of the international conference ‘Artificial Intelligence and Social Research: innovative approaches for the analysis of forms of antagonism’, which brings together international experts for two days at the Public University of Navarra (UPNA) to analyze the role of technology in current challenges, explained it in statements to journalists.
The expert has warned of the growing impact of the so-called “normative antagonism”—which includes disinformation, polarization, xenophobia or hate speech—in current democracies, and has presented the “Xenometer” tool as an instrument to analyze these phenomena on social networks.
The conference also seeks to make visible the social impact of polarization, a phenomenon that, according to the expert, “not only affects the United States, but in Europe is putting social cohesion at risk.”
In this context, the researcher has highlighted the role of “Xenometer”, a tool initially developed at Cornell University (United States) during Donald Trump’s first term, given the concern about the spread of xenophobic narratives on social networks.
The system allows messages about migration to be classified based on their tone—positive or negative—and their intensity on a seven-level scale, so it not only detects xenophobia, but also grades it, which allows for a more concrete analysis, he explained.
He has detailed that the tool achieves 97% effectiveness in identifying whether a message is about migration and 95% in determining whether its focus is positive or negative. However, it still presents difficulties in distinguishing more subtle nuances, such as irony or the intensity of the speech.
The project, which has been carried out in Spain since 2023, has already analyzed nearly 5,000 messages, although these do not constitute a representative sample, since they have been strategically selected to train the algorithm.
One of the main challenges detected has been the influence of the cultural context on the interpretation of messages, since “we realized that the subtleties linked to humor and irony are not captured unless you are very involved in a culture,” he noted, which has led to the creation of national teams in different countries.
The final objective is to develop a “world map of digital hate”, through an international observatory in which the universities currently involved participate and prepare semi-annual reports on the evolution of these discourses on social networks.
Beyond academic analysis, García-Magariño has highlighted the possible practical applications of “Xenometer”, such as its use by security forces to detect hate crimes or by public administrations to improve content moderation on digital platforms. EFE
(Photo)