Download presentation
Presentation is loading. Please wait.
1
Computer Science Department, Peking University
VoteTrust Leveraging Friend Invitation Graph to Defend Social Network Sybils Jilong Xue , Zhi Yang , Xiaoyong Yang, Xiao Wang, Lijiang Chen and Yafei Dai Computer Science Department, Peking University
2
Sybil attack in Social networks
Sybils Friend invitation Non-popular users reject accept
3
VoteTrust: An Overview
Basic idea: Considering invitation feedback as voting Key techniques: Trust-based votes assignment Global vote aggregation Properties: High precision in Sybil detection Efficient in limiting Sybilβs attack ability
4
Graph Model A A 1 C C B B Link initiation graph πΊ πΌ =( π πΌ , πΈ πΌ )
B B Link initiation graph πΊ πΌ =( π πΌ , πΈ πΌ ) Link acceptance graph πΊ πΈ =( π πΈ , πΈ πΈ , π πΈ )
5
Framework of VoteTrust
Select trust seed β high reliable users Distribute votes Collect votes and computing score
6
Outline Preliminary Implementation Evaluation Conclusion
Trust-based vote assignment Global vote aggregation Evaluation Conclusion
7
Votes Assignment Problem: Principle: How to implement?
How to distribute votes across users? Principle: Reliable user should get more votes How to implement? v v Reliable user v v v Non-popular user Sybil
8
Trust-based Votes Assignment
Step1: Assigning votes to little human-selected reliable seeds Step2: Propagating to whole users across the Link initiation graph ππππ(π’)=πβ π£:(π£,π’)β πΈ πΌ ππππ(π£) ππ’π‘_ππππππ(π£) + 1βπ βππππ(π’)
9
Example β¦ A B C D E A B C D E Node A is reliable seed Total votes =5
t=1 0.75 4.25 t=2 0.75 0.65 1.80 1.80 C D E t=3 2.59 0.94 0.31 0.58 0.58 β¦ t=n 1.69 1.57 0.14 0.80 0.80 Node A is reliable seed Total votes =5
10
Outline Preliminary Implementation Evaluation Conclusion
Trust-based vote assignment Global vote aggregation Evaluation Conclusion
11
Vote Aggregating Problem: Principle:
How to collect votes and compute user trust score? Trust score π π’ β[0,1] Principle: Trust user should have high weight in voting. vote=1,score=0.2 vote=1,score=0.9 A B 1 C score=? π ππππ(πΆ)= 1Γ0.9 1Γ0.2+1Γ0.9 =0.82
12
Global Vote Aggregation
Step1: Set all usersβ initial score as 0.5; Step2: Iteratively computing each userβs trust score according to aggregated votes. πππππ π’ = π£ππ‘π π£ βπππππ π£ β π₯ π£,π’ π£ππ‘π π£ βπππππ π£ , (π£,π’)β πΈ πΈ
13
Small-sample Problem Number of votes is too small. Wilson score
weighted average of π and vote=1,score=0.2 A B score=0 ? score=0.40 π= π + 1 2π π§ 1β πΌ π π§ 1β πΌ 2 vote=1,score=0.2 A 1 B score=1 ? score=0.61
14
Security Properties (I)
Theorem 1: The number of Sybilβs attack-link needs to satisfy the following upper bound where πΏ π is detection threshold. π΅ πππ β€π π΅ ππ β πΏ π β πΏ π 2 πΏ π βπ π ππ Normal user Sybil π ππ’π‘
15
Simulation of Theorem 1 Comm size: 100 # of in-links: 10
Nout avg: 2.36 Nout max:4
16
Security Properties (II)
Theorem 2: Sybil community size need to satisfy the upper bound , where πΏ π£ is vote collection threshold. π΅ π β€πβ π΅ ππ πΏ π£
17
Simulation of Theorem 2
18
Outline Preliminary Implementation Evaluation Conclusion
Trust-based vote assignment Global vote aggregation Evaluation Conclusion
19
Experimental Setup Data Set Methodology
Renren regional network (PKU) include 200K users, million friend invitations 2502 Sybil accounts detected by Renren Manual checking 73 Sybils from 500 random user Methodology Compared with TrustRank and BadRank Evaluation metrics: Precision and Recall
20
TrustRank vs. VoteTrust
Averagely improve 32.9% Averagely improve 75.6%
21
BadRank vs. VoteTrust Averagely improve 44.5% Averagely improve 41.6%
22
Separating Normal User from Sybils
80% with low score
23
Separating Normal User from Sybils
Maximum accuracy=85.7%
24
Performance Summary Outperforms TrustRank by 32.9% in detection precision averagely; Outperforms BadRank by 44.5% in detection precision averagely; High accurate in classifying the Sybil and normal user (include non-popular user)
25
Outline Preliminary Implementation Evaluation Conclusion
Trust-based vote assignment Global vote aggregation Evaluation Conclusion
26
Conclusion VoteTrust is a rating system Key techniques
high accuracy in Sybil detection Efficient in resisting Sybil (community) Key techniques Trust-based vote assignment Global vote aggregation
27
Thank you!
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.