Skip to main content
Fig. 21 | Computational Social Networks

Fig. 21

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

Fig. 21

Increasing link-prediction error under linear threshold model: \(\text {OAS}_{\text {mg}}\) in UCI messenger-app network (1281 nodes). Linear threshold model on the UCI messenger-app network with (right panel) or without (left panel) information on realized node thresholds. Largest budget (top panels) is sufficient for a full cascade under greedy seeding when realized thresholds and perfect link information are known. Half this budget and a quarter of this budget are also tested (panels labeled). Note the variable scales on the vertical axes. Mean random-seeding performance is shown in red

Back to article page