Tie strength in co-authorship social networks: analyses, metrics and a new computational model

Abstract

In social networks, tie strength has been largely explored in different contexts, such as information diffusion and evaluation of scientific productivity. In this thesis, we measure the strength of co-authorship ties in non-temporal and temporal social networks. In sum, the main contributions are: (1) a survey and a taxonomy of social professional networks; (2) an analysis of how topological properties relate to tie strength in non-temporal networks, which reveal how different properties explain variations in tie strength; (3) a new metric called tieness that better differentiates tie strength in distinct levels in non-temporal networks; (4) a new algorithm to measure tie strength called STACY (Strength of Ties Automatic-Classifier over the Years), which is able of better identifying strong ties than a state of the art algorithm; and (5) a new computational model called temporal_tieness to classify tie strength in temporal networks with low computational cost.