Measuring Users Relationship Strength Using a Hierarchical Voting-Based Model

In this study, we proposed a user relationship strength hierarchy vote (URSHV) model to measure the relationship strength, which can reflect the degree of intimacy between two different persons. This model is a hierarchical model measuring the relationship by voting on users' daily moving trajectory similarity, physical position similarity,, the corresponding semantic location similarity, capable for both GPS, base station data. The daily trajectory similarity is measured using the dynamic time warping model, then weighted by the entropy of the trajectory sequence. The similarities of physical position, semantic location are calculated through the latent Dirichlet allocation topic model. Ultimately, we voted on these three similarities for the final relationship strength. The proposed model was evaluated using an open data set, the experiments showed its validity in inferring users relationship strength.