Modelagem automática de perfis de usuários do Twitter utilizando diferentes técnicas de enriquecimento semântico

• Ano de defesa: 2016
• Autor(a) principal: Mendes, Diego Sarmento
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Lavras; Programa de Pós-Graduação em Ciência da Computação; UFLA; brasil; Departamento de Ciência da Computação
• Palavras-chave em Português: Redes sociais; Usuários da Internet – Análise; Semântica – Análise de redes; Mineração de dados; Social networks; Internet users – Analysis; Semantics – Networks analysis; Data mining; Twitter; Ciência da Computação
• Link de acesso: http://repositorio.ufla.br/jspui/handle/1/12324

Increasingly, social networking sites stands out by the large volume of information created by users daily, which are composed of text, photos and videos. However, it is still a big challenge to use user’s published data to understand precisely their interests, which is a valuable information in many applications. To deal with such difficulties, advanced techniques to add semantic value to the texts were proposed in other studies, obtaining implicit information that are present in their own publications or news URLs mentioned by users. Following this same idea, in this research we propose a new approach of semantic enrichment for Twitter users’ publications, which it is considered information that is beyond the textual content present in publications, also exploring the extracted concepts from images in publications and news shared by users. Therefore, the main contribution of this work is to create an automatic modeling tool for Twitter user’s profiles, using different state-of-the-art techniques to semantic enrichment based on textual content, as well as the proposed new approach using images, comparing them in scenarios real involving a news recommendation system and classification of tweets, respectively. Moreover, an application to show the created profiles, and their respective changes over time was implemented, allowing data to be collected and enriched in real time.