Creating Artificial Intelligence (AI) models capable of understanding human language is crucial for the development of several applications based on natural language processing (PLN) – and, also, a great challenge, especially if these models are applied to very subjective scenarios. , such as analysis of feelings, detection of hate speech or irony. This is the focus of the academic competition Irony Detection in Portuguese – IDPT 2021, which started in late March and has the participation of a team of experts in AI Intelligence from SiDi.
The competition is scheduled for a month and a half and is organized by the Federal University of Pelotas and is part of the Iberian Languages Evaluation Forum – IberLEF 2021, which aims to encourage the research community to promote competitive tasks in the area of information processing systems. natural language in Spanish and other Iberian languages. In the case of IDPT, the challenge will be to identify the presence of irony in news texts and tweets written in Portuguese.
“The competition will be all virtual and, during its realization, we will receive several texts that will be used to train specific AI models”, explains Fernando Asevedo, who is part of the SiDi PLN team. “In another stage, the organization of the event will send the tests to apply the models developed. We will then have a deadline to send the answers, based on which the organization will disclose the final score of the competition ”, he adds.
According to Asevedo, participation in competitions of this type is a form of training for professionals who work in constantly evolving areas, such as natural language processing. “In addition, SiDi’s closer relationship with the academic community contributes to attracting new talent to our teams”, he concludes.