Informação de links no modelo vetorial usando uma estrutura funcional

Ilmério Reis SilvaJoão Nunes de SouzaRenata Ferreira Lisboa MouraBerthier Ribeiro-Neto

The Classical Models of Information Retrieval (IR) only consider the content-based information to retrieve relevant documents. The information is captured by comparing document keywords against the user query terms. Considering the Web, the structure of links among the documents (or pages) is a rich source of information that has not been used by the Classic Models. On the other hands, Bayesian Network models for IR has revealed that link-based combined with content-based information can improve the quality of the retrieved documents. In this work we discuss this combination in the context of the Classic Vector Space Model(VSM). We present three new contributions. We make the combination of link-based with contentbased information close to the IR community, through an extension of the VSM. We define a query language that allows users to state document preferences according to the structure of links. Finally, we propose a metric that allows the comparison among documents using link-based information.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: