DEFESA DE PROPOSTA DE TESE
Aluna: Brenda Salenave Santana
Orientador: Prof. Dr. Leandro Krug Wives
Título: A Linguistic-computational-based Approach to Support the Analysis of the Discursive Configuration of Violence on Social Media
Linha de Pesquisa: Mineração, Integração e Análise de Dados
Esta banca ocorrerá excepcionalmente de forma totalmente remota. Interessados em assistir a defesa poderão acessar a sala virtual através do link: http://mconf.ufrgs.br/webconf/00108957.
Profª. Drª. Raquel da Cunha Recuero (UFPEL)
Profª. Drª. Renata de Matos Galante (UFRGS)
Prof. Dr. Thiago Pardo (USP)
Presidente da Banca: Prof. Dr. Leandro Krug Wives
Abstract: Researches aimed at studying hate speech have grown in recent years; however, approaches capable of automatically detecting this type of content still have significant limitations. These limitations are even more latent in languages with scarce data, such as Portuguese. This work proposes to study the use of linguistic indicators characteristic of hate speech associated with computational methods (such as the classification of texts by machine learning algorithms). It seeks to evaluate the use of such indicators to enable the framing of content disseminated on social networks from the perspective of Frame Semantics, considering the instantiation of a frame of symbolic violence as a way of proposing a means of interposing intolerant speeches. As a primary source of data, we consider Twitter. The data extraction will be performed encompassing conditions considered relevant for our research, which present topics considered to be potentially bearers of hate speech. Among the themes to be considered in this study, this research seeks to focus mainly on topics related to political gender violence. When dealing with the data to be observed, it is proposed to rank texts collected through Perspective API features as Identity Attack, Insult, Profanity, Toxicity, and Threat. To carry out the analysis of the data involved in the research, a quanti-quali approach will be considered. From the qualitative aspect, at first, content analysis will be performed using computational approaches (use of the Perspective API for data selection). It will serve as a basis for a qualitative analysis of the observations and the questions opened in a questionnaire. On the other hand, the quantitative approach will make use of statistical analyses, such as frequency distributions, correlations, and graphical representations, measures of dispersion, measures of central tendency, thus seeking to observe how the present proposal fits or not with the expected outcomes. Thus, it is expected to contribute to academic study development centered on social media, focusing on potentially harmful speeches written in Portuguese, also taking into account and the user’s perception of the content generated on social networks and their repercussions in daily life.
Keywords: Symbolic Violence; Hate Speech; Intolerant Speech; Frame Semantics.