Skip to main content
Fig. 7 | Human-centric Computing and Information Sciences

Fig. 7

From: Trust reality-mining: evidencing the role of friendship for trust diffusion

Fig. 7

Simulation results for \(\alpha = 0.1\) and \(\beta = 0.1\) for \(K_I\) and \(K_C\). Figure a and b show the f-test results for the distributions of (a) our trust model (TM) versus the real data and (b) a random model (RM) versus the read data. In both cases p is large indicating the model distributions are adequately fitting the data and are not statistically significantly different. Note, the same experiments with friendship replaced by political discussants produced statistically significantly differing fits. The quantile-quantile (QQ plot) for (c) the TM versus the real data and (d) the RM versus the real data. The better the distribution fits the real data, the more the slope approaches 1. The slopes are shown in Table (e), along with other measures comparing the distributions. Considering all 3 measures, our trust model better fits the real data at the end of the academic year (June 2009) in comparison to a well performing random model used as a baseline

Back to article page