Objects description exploiting user's sociality

Abstract

Declarative scene modelling is a very useful modelling technique which allows the user to create scenes by simply describing their wished properties and not the manner to construct them. In declarative modelling, solution filtering is a very important aspect due to the imprecise description of both the scene and the objects, as well as due to the subjective of humans regarding the content of a design is concerned. Currently, solution filtering is performed by the application of machine learning strategies or clustering methods in a collaborative or not framework. However, the main difficulty of these algorithms is that solution filtering is based on the usage of low-level attributes that describe either the scene or the object. This chapter addresses this difficulty by proposing a novel social oriented framework for solution reduction in a declarative modelling approach. In this case, we introduce semantic information in the organization of the users that participates in the filtering of the solutions. Algorithms derived from graph theory are presented with the aim to detect the most influent user with a social network (intra-social influence) or within different social groups (inter-social influence). Experimental results indicate the outperformance of the proposed social networking declarative modelling with respect to other methods. © 2012 Springer-Verlag Berlin Heidelberg.