STACY: Strength of Ties Automatic-Classifier over the Years

Abstract

With the evolution of Web technology and its worldwide use by regular people, there is now data about not only such people but also their relations. Database research has evolved as well to tackle the myriad of problems that arrive with such volumes of data. Here, we contribute to such a trend by proposing a new algorithm (STACY) to automatically classify tie strength (an intrinsic property of relationships) considering time. We show that each class has singular and different behavior, and analyze them over co-authorship networks. Also, STACY identifies strong relationships that persist more than the ones classified by a state of the art algorithm. Finally, we derive a computational model from STACY that is able to automatically identify relationships classes with low computational cost.