The analysis of online hate speech in Portugal was highlighted at the final conference, held on July 9 at Iscte, of the kNOwHATE project, funded by the European Union.
The research team presented the main results of the project and innovative tools and resources to understand and combat this problem.
One of the main conclusions of the study is that there is not one, single, hate speech, but rather multiple hateful discourses, which mobilize different contents and use diverse linguistic-discursive strategies, with the most veiled and subtle forms of expression being the most prevalent in the data analyzed. According to Rita Guerra, the project's coordinator, "hate is hidden in seemingly harmless messages, that can carry great potential to cause harm".
To help tackle this problem, the kNOwHATE researchers developed several resources. One of them is a prototype that uses AI to automatically classify hate speech in texts written in Portuguese. This prototype is available in open access platform, and although it is still under development and may contain errors, it offers users the chance to test an automatic detection system and understand how it works.
"This tool can be used in educational contexts, but essentially the main target audience is the “silent majority”, the ordinary citizen who doesn't necessarily have an engineering background and who may not even know what an algorithm is," said Rita Guerra, the project's coordinator.
The fight against hate speech is a challenge that requires everyone's collaboration and the use of multi-level responses. Effectively combating this phenomenon cannot be done simply by moderating online content and using automatic detection mechanisms. Legal, educational, artistic and awareness-raising responses to this phenomenon are needed. To this end, the project developed the campaign #CortarOMalPelaRaiz, which includes various online resources, a 4-episode podcast, a brochure, a flyer and a video for social media. It is crucial that digital platforms, governments, schools and civil society work together to create a safer and more inclusive online environment. The KNOWHATE project has taken an important step in this direction, but there is still a long way to go.
The conference program also featured the keynote speaker Sahana Udupa, professor of Media Anthropology at LMU University of Munich and a specialist in hate speech, who warned of the enormous limitations of current content moderation models based on machine learning, highlighting the need to invest in transparent, reflective and collaborative approaches to content moderation.
Also present were the kNOwHATE project's civil society partners, SOS Racismo, Casa do Brasil de Lisboa and ILGA.
Víctor Costa, representative of Casa do Brasil de Lisboa, stressed the importance of collaboration between academia and civil society, saying that "it's good to see that academia and civil society are mobilizing to try to understand this phenomenon and are organizing themselves to create moderation mechanisms." He also pointed out that "hate speech is starting to leave the online space and reach common spaces, workplaces and social spaces."
Sara Sores, representing ILGA, stressed her concern about the increase in hate speech, especially during the pandemic. According to the association's representative, hate speech is becoming more and more present in our spaces and at our events, leaving the online environment and manifesting itself in an increasingly concrete way.
José Falcão, representing SOS Racismo, highlighted the seriousness of some recent attacks against migrants as worrying examples of the link between what happens online and offline, stressing the need to invest in appropriate legal responses and education as means of combating the proliferation of racism and xenophobia.
