Multiresolution organization of social media users' profiles: fast detection and efficient transmission of characteristic profiles

Abstract

In this paper, multiresolution organization of social media users' profiles is proposed that enables fast browsing and effective transmission of characteristic profiles. The scheme results in the construction of a novel three-layer tree structure. At layer 0 the root-node contains the most representative profile among class representatives (CRs). Each node of layer 1 represents a class of profiles, while in layer 2 the full resolution is available (all profiles). The nodes of the proposed tree structure contain viewing elements and in this paper we focus on the extraction of viewing elements for layers 0, and 1. Towards this direction the viewing elements of layer 1 are optimally extracted using an interpolation method, while the viewing element of layer 0 is estimated from the viewing elements of layer 1. The resulting tree-structure enables users to quickly browse and detect profiles of interest, by selecting the viewing elements they like. Experimental results on reallife social media users indicate the promising performance of this innovative scheme.