Ontologias para representação de documentos multimídia: análise e modelagem

• Ano de defesa: 2014
• Autor(a) principal: Daniela Lucas da Silva
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Representação do conhecimento; Anotação semântica; Ontologias; Padrões de metadados; Ontologias multimídia; Web Semântica; Metadados; Ciência da informação; Representação do conhecimento (Teoria da informação); Ontologias (Recuperação da informação)
• Link de acesso: http://hdl.handle.net/1843/BUOS-9NCGYM

At the global level, it is evident the significant growth of semantically related data on the Web. Metadata standards recommended by the World Wide Web Consortium (W3C) have been used to describe and represent multimedia resources, enabling and expanding the access and improving the management, organization and retrieval of digital archives. A common issue in institutions that make use of digital archives of various natures is the integrated approach to heterogeneous databases and the lack of standardization in the description formats. The description of the items is usually performed independently, with idiosyncratic patterns of description, highlighting different features to be described and using different terminology to describe them. This thesis aimed at advancing the research state-ofart on semantic representation of documents in not only textual formats, but also involving multimedia content, as videos, images and audio; proposing a reference multimedia conceptual model, that allow for the systematic organization of typologies of existing metadata for describing multimedia content across a variety of contexts and needs. The modeling approach was methodologically based on the best multimedia ontologies, evaluated in regard to the reuse, which ensured the selection of appropriate knowledge resources for the integration in the conceptualization model proposed. This model sought covering functional and non-functional requirements determined in the research, proposing fundamental ontological classes and relationships arising from the combination and alignment of previously analyzed and evaluated multimedia ontologies, providing a comprehensive architecture for the semantic organization of multimedia metadata, addressed mainly to applications that deal with information resources on the Web.