Current models of ranking in Information Retrieval (IR) are somewhat blind to the consideration of the social context that surrounds an information resource as a parameter that affects the precision of the ranking in the query results. On the other hand that social context is depicted upon the relational ties of the affiliated social entities (authors) thus is something that cannot be measured and quantiﬁed accurately by back-link and citation based models. In this thesis we adopt an imprecise modeling approach of the depicted relational ties using the paradigm of fuzzy sets as to express partial degrees of membership depicted on the concept of 'opinion' as an input to a model that considers both the informational (hyperlink) and social (relational) context of the information resources as to provide better ranking of the retrieved results with respect to both contexts. A formalization of the algorithm and a validation of the model using simulation is given as a proof of concept along with discussion of the obtained results.
|Media of output
|Royal Institute of Technology (KTH), Stockholm
|Place of Publication
|Published - 2005