ComentCorpus: identificação e pistas linguísticas para detecção de ironia no português do Brasil

• Ano de defesa: 2018
• Autor(a) principal: Pedro, Gabriela Wick
• Orientador(a): Vale, Oto Araújo
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de São Carlos; Câmpus São Carlos
• Programa de Pós-Graduação: Programa de Pós-Graduação em Linguística - PPGL
• Palavras-chave em Português: Análise de sentimentos; Linguística de corpus; Ironia; Opinião pública; Processamento de linguagem natural (Computação)
• Palavras-chave em Inglês: Sentiment analysis; Public opinion; Natural language processing (Computer science); Irony; Corpora (Linguistics)
• Área do conhecimento CNPq: LINGUISTICA, LETRAS E ARTES::LINGUISTICA::LINGUISTICA APLICADA
• Link de acesso: https://repositorio.ufscar.br/handle/20.500.14289/10710

Opinions on the Web have been increasing progressively and, thus, has aroused interest in areas of study of Linguistics and Computation, for example. In this context comes the Sentiment Analysis, or Opinion Mining, which aims to analyze computationally opinions, emotions, feelings and subjectivities present in texts (LIU, 2012), however, certain subjective sentences can carry irony, transforming the meaning of a sentence. This dissertation aims to investigate expressions of irony in social media, focusing on the description of linguistic devices as clues of irony in opinion texts in Brazilian Portuguese. To understand the functioning of this figurative mechanism, we will start from the search a corpus constructed by new commentaries from the Folha de S. Paulo portal. In addiction, based on pragmatic and cognitive theories, we developed a corpus annotation scheme for opinions and their intentions: ironic, other types of irony or non-ironic. As a result, we have obtained a list of subcategories that characterize expressions of irony that allow to collaborate with the development of the NLP area and Sentiment Analysis and, in addition, to improve tools of automatic identification of opinion through the descriptions and the linguistic resources elaborated here.