Duration: 1 December 2022 to 30 November 2024
Grant PDC2022-133118-I00 funded by MCIN/AEI/10.13039/501100011033 and by European Union NextGenerationEU/PRTR
PI: Paolo Rosso
Members: José Miguel Benedí, Jon Ander Gómez, Roberto Paredes, Alejandro H. Toselli

Social media have become the default channel for people to access information and express ideas and opinions. A perverse effect is that social media are a breeding ground for the propagation of fake news: when a piece of news outrages us or matches our beliefs, we tend to share it without checking its veracity; and, on the other hand, content selection algorithms in social media give credit to this type of popularity because of the click-based economy on which their business are based. Another harmful effect is that the relative anonymity of social networks facilitates the propagation of hate speech and exclusion messages. Therefore, social media may contribute, paradoxically, to the disinformation spread and polarization of society.
In the framework of the MISMIS-FAKEnHATE project (PGC2018-096212-B-C31) we developed software for FAKE news and HATE
speech. We addressed the problem of FAKE news detection taking into account the emotions that fake news may trigger in the readers, and psycholinguistics aspects such as the personality traits that we empirically proved that helped in discriminating between fake news spreaders and fact-checkers. Moreover, we addressed it from a multimodal perspective taking into account also the information contained in images. We registered our software at the Office for the Promotion of Research, Innovation and Technology Transfer of the UPV in order to ease the transfer of the research results. Our developed software has been validated in lab and has a technology readiness level of 4. We have been collaborating with Symanto Spain also in the framework of a project on the detection of conspiracy theories. A proof of concept in Symantos industrial environment should allow us to reach TRL 6.
With respect to the detection of HATE speech, our major efforts have been in the identification of implicit hate speech with stereotypes against immigrants. We developed transformer-based software that has been validated in lab on the StereoImmigrants dataset composed of Spanish politicians’ speeches. Since the end of 2021 we have been collaborating with OBERAXE, the Spanish observatory on racism and xenophobia of the Secretary of State for Migration (Public Procurement File 2021/30000225 ending at the end of May 2022).
Our goal has been to provide technical and logistical support service for monitoring hate speech online (in Twitter, Facebook, Instagram, YouTube, and TikTok), and concretely for to develop a data collection application that will facilitate to compile a dataset of hate speech cases of social media. Collaborating with OBERAXE also in the framework of the FAKEnHATE-PdC project will allow us to train our software for hate speech and stereotype identification with social media data on the basis of the taxonomy we proposed. In the OBERAXEs operational environment, we aim to reach a technology readiness level of 7, equivalent to the system prototype demonstration.