Skip to main content
Fig. 5 | Computational Social Networks

Fig. 5

From: Measuring the value of accurate link prediction for network seeding

Fig. 5

Increasing link-prediction error in a small-world network (larger communities, \(\text {OAS}_{\text {tg}}\)). The experiment from Fig. 4 is replicated with the traditional greedy algorithm. \(\text {OAS}_{\text {tg}}\) as a function of false negative rate in a small-world network (rewiring probability of 0.4, mean community size of 20, standard deviation 5). Traditional greedy allocation of \(b\) seeds causes a complete cascade in \(G'\) (300 nodes). Budgets are (10, 68, 153, 235). At all but the lowest threshold, budgets required by traditional greedy in \(G'\) to achieve full cascades are much higher than those required by modified greedy: the average performance of a random seed set in \(G\) (shown in red) is higher for each subplot than in Fig. 4

Back to article page