IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING .2 料料 6 .6 Classic Greedy Classie Greedy -D小鱼i长Sehc组a 100 60 120 20 ”” 110 (a)Facebook,BlkPer=1% (b)Facebook,BlkPer=2% (c)Facebook,BlkPer=10% 4168888885 0.3 0.8 0.7 Normal Preps2a —Normal Propagation -Classie Greedy -出ssie Greed 3 21 30 ”me 100 12n 20 100 (d)Twitter,BlkPer=1% (e)Twitter,BlkPer=2% (f)Twitter,BlkPer=10% 03 -25 25 日 0.2 a.1 -D 6 10 20 0 100 120 100 114 (g)Sina,BlkPer=1% (h)Sina,BlkPer=2% (i)Sina,BlkPer=10% Fig.4.The experimental results of the rumor infection ratio with propagation iterations under different blocking algorithms in the Facebook ((a),(b),(c)),Twitter((d),(e),(f))and Sina Weibo ((g),(h),(i))dataset respectively.The blocking percentage of all the nodes in the social network is set to 1%,2%and 10%for each dataset. lassie G 0.2 0.14 012 022 1.013 0.2 .2 .01 0202 (a)Facebook (b)Twitter (c)Sina Fig.5.Stationary rumor infection ratio under different blocking algorithms with different blocking ratios on the Facebook,Twitter and Sina Weibo datasets respectively.The blocking ratio ranges from 1%to2%with an interval of 0.1%,which shows the sensitivity of different blocking algorithms. TABLE 2 False Positive Ratio in Nodes Blocking Data Sets Facebook Twitter Sina Weibo Classic Greedy 42% 40% 70% Proposed Greedy 34% 36% 64% Dynamic Schema 14% 18% 34%IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING 10 0 20 40 60 80 100 120 0 0.2 0.4 0.6 0.8 1 1.2 Time (iterations) Infection Ratio Normal Propagation Classic Greedy Proposed Greedy Dynamic Schema (a) Facebook, BlkPer=1% 0 20 40 60 80 100 120 0 0.2 0.4 0.6 0.8 1 1.2 Time (iterations) Infection Ratio Normal Propagation Classic Greedy Proposed Greedy Dynamic Schema (b) Facebook, BlkPer=2% 0 20 40 60 80 100 120 0 0.2 0.4 0.6 0.8 1 1.2 Time (iterations) Infection Ratio Normal Propagation Classic Greedy Proposed Greedy Dynamic Schema (c) Facebook, BlkPer=10% 0 20 40 60 80 100 120 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Time (iterations) Infection Ratio Normal Propagation Classic Greedy Proposed Greedy Dynamic Schema (d) Twitter, BlkPer=1% 0 20 40 60 80 100 120 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Time (iterations) Infection Ratio Normal Propagation Classic Greedy Proposed Greedy Dynamic Schema (e) Twitter, BlkPer=2% 0 20 40 60 80 100 120 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Time (iterations) Infection Ratio Normal Propagation Classic Greedy Proposed Greedy Dynamic Schema (f) Twitter, BlkPer=10% 0 20 40 60 80 100 120 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 Time (iterations) Infection Ratio Normal Propagation Classic Greedy Proposed Greedy Dynamic Schema (g) Sina, BlkPer=1% 0 20 40 60 80 100 120 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 Time (iterations) Infection Ratio Normal Propagation Classic Greedy Proposed Greedy Dynamic Schema (h) Sina, BlkPer=2% 0 20 40 60 80 100 120 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 Time (iterations) Infection Ratio Normal Propagation Classic Greedy Proposed Greedy Dynamic Schema (i) Sina, BlkPer=10% Fig. 4. The experimental results of the rumor infection ratio with propagation iterations under different blocking algorithms in the Facebook ((a),(b),(c)), Twitter ((d),(e),(f)) and Sina Weibo ((g),(h),(i)) dataset respectively. The blocking percentage of all the nodes in the social network is set to 1%, 2% and 10% for each dataset. 0.01 0.012 0.014 0.016 0.018 0.02 0.022 0 0.05 0.1 0.15 0.2 0.25 Blocked Nodes Ratio Rumor Affected Nodes Ratio Normal Propagation Classic Greedy Proposed Greedy Dynamic Schema (a) Facebook 0.01 0.012 0.014 0.016 0.018 0.02 0.022 0 0.05 0.1 0.15 0.2 Blocked Nodes Ratio Rumor Affected Nodes Ratio Normal Propagation Classic Greedy Proposed Greedy Dynamic Schema (b) Twitter 0.01 0.012 0.014 0.016 0.018 0.02 0.022 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 Blocked Nodes Ratio Rumor Affected Nodes Ratio Normal Propagation Classic Greedy Proposed Greedy Dynamic Schema (c) Sina Fig. 5. Stationary rumor infection ratio under different blocking algorithms with different blocking ratios on the Facebook, Twitter and Sina Weibo datasets respectively. The blocking ratio ranges from 1% to 2% with an interval of 0.1%, which shows the sensitivity of different blocking algorithms. TABLE 2 False Positive Ratio in Nodes Blocking Data Sets Facebook Twitter Sina Weibo Classic Greedy 42% 40% 70% Proposed Greedy 34% 36% 64% Dynamic Schema 14% 18% 34%