Figures and tables with the experimental results related to the work "Exploiting Social and Mobility Patterns for Friendship Prediction in Location-Based Social Networks", accepted to be published in proceedings of The 23rd International Conference on Pattern Recognition (ICPR 2016).
· Table 1: Performance comparison of different methods for inferring social links on unsupervised domain.
· Table 2: Classifier results measured by AUC.
· Figure 1: Number of correctly and wrongly predicted links for social (G1), location (G2), and proposed (G3) methods for (a) Brightkite, and (b) Gowalla. The dashed horizontal line indicates the number of truly new links (links into the probe set). Results averaged over the 10 analyzed partitions.
· Figure 2 Nemenyi post-hoc test diagrams obtained from (a) unsupervised and (b) supervised experiment results showed in Tables 1 and 2, respectively.
· Figure 3 Precisi@L performance of the top five methods considering different L values for (a) Brightkite, and (b) Gowalla.