Complex Interactions in Social and Event Network Analysis

Abstract

Modern social network analytic techniques, such as centrality analysis, outlier detection, and/or segmentation, are limited in that they typically only identify interactions within the dataset occurring as a first-order effect. In our previous work, we illustrated how the use of tensor decomposition can be used to identify multi-way interactions in both sparse and dense data-sets. The primary aim of this paper will be to introduce innovative extensions to our tensor decomposition approach that target and/or identify second and third order effects.