VoteTrust Leveraging Friend Invitation Graph to Defend Social Network Sybils

27
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

description

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. Sybil attack in Social networks. Sybils. Friend invitation. reject. - PowerPoint PPT Presentation

Transcript of VoteTrust Leveraging Friend Invitation Graph to Defend Social Network Sybils

Page 1: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

VoteTrustLeveraging 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

Page 2: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Sybil attack in Social networks

Non-popular users

Sybils

Friend invitation

reject

accept

Page 3: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

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 Sybil’s attack ability

Page 4: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Graph Model

A

B

C

A

B

C

Link initiation graph Link acceptance graph

1

0

Page 5: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Framework of VoteTrust

• Select trust seed – high reliable users• Distribute votes• Collect votes and computing score

Page 6: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Outline

PreliminaryImplementation

– Trust-based vote assignment– Global vote aggregation

EvaluationConclusion

Page 7: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Votes Assignment

• Problem: – How to distribute votes

across users?

• Principle:– Reliable user should get

more votes

• How to implement?

vvvvvReliable user

Non-popular user Sybil

Page 8: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Trust-based Votes Assignment

• Step1: Assigning votes to little human-selected reliable seeds

• Step2: Propagating to whole users across the Link initiation graph

𝒗𝒐𝒕𝒆(𝑢)=𝑑 ∙ ∑𝑣: (𝑣 ,𝑢)∈𝐸 𝐼

❑ 𝒗𝒐𝒕𝒆(𝑣)𝑜𝑢𝑡𝑑𝑒𝑔𝑟𝑒𝑒(𝑣)

+(1 −𝑑 ) ∙ 𝒊𝒏𝒊𝒕 (𝑢)

Page 9: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Example

A B

DC E

Node A is reliable seed Total votes =5

A B C D E

5 0 0 0 0

0.75 4.25 0 0 0

0.75 0.65 0 1.80 1.80

2.59 0.94 0.31 0.58 0.58

1.69 1.57 0.14 0.80 0.80

…t=0

t=1

t=2

t=3

t=n

Page 10: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Outline

PreliminaryImplementation

– Trust-based vote assignment– Global vote aggregation

EvaluationConclusion

Page 11: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Vote Aggregating

• Problem:– How to collect votes and

compute user trust score?– Trust score

• Principle:– Trust user should have

high weight in voting.

A B

C

0 1

vote=1,score=0.2 vote=1,score=0.9

score=?

Page 12: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Global Vote Aggregation

• Step1: Set all users’ initial score as 0.5;• Step2: Iteratively computing each user’s trust

score according to aggregated votes.

𝒔𝒄𝒐𝒓𝒆 (𝑢)=∑ 𝑣𝑜𝑡𝑒 (𝑣 ) ∙𝒔𝒄𝒐𝒓𝒆 (𝑣 ) ∙𝑥𝑣 ,𝑢

∑ 𝑣𝑜𝑡𝑒 (𝑣 ) ∙ 𝒔𝒄𝒐𝒓𝒆 (𝑣 ),(𝑣 ,𝑢)∈𝐸𝐸

Page 13: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Small-sample Problem

• Number of votes is too small.

• Wilson score

– weighted average of and .

A B

vote=1,score=0.2

0 score=0 ?

A B

vote=1,score=0.2

1 score=1 ?𝒑=�̂�+

12𝑁

𝑧1 −𝛼/2❑

1+ 1𝑁

𝑧1 −𝛼/2❑

score=0.40

score=0.61

Page 14: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Security Properties (I)

• Theorem 1: The number of Sybil’s attack-link needs to satisfy the following upper bound

where is detection threshold.

𝑵𝒐𝒖𝒕 ≤𝜌𝑵 𝒊𝒏 ∙𝛿 𝑓 −𝛿 𝑓

2

𝛿 𝑓 −𝑟

𝑁 𝑖𝑛

𝑁 𝑜𝑢𝑡

Normal user Sybil

Page 15: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Simulation of Theorem 1

• Comm size: 100• # of in-links: 10

• Nout avg: 2.36• Nout max:4

Page 16: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Security Properties (II)

• Theorem 2: Sybil community size need to satisfy the upper bound ,

where is vote collection threshold.

𝑵 𝒔≤𝜎 ∙𝑵 𝒊𝒏

𝛿𝑣

Page 17: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Simulation of Theorem 2

Page 18: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Outline

PreliminaryImplementation

– Trust-based vote assignment– Global vote aggregation

EvaluationConclusion

Page 19: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Experimental Setup

• Data Set– Renren regional network (PKU) include 200K

users, 5.01 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

Page 20: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

TrustRank vs. VoteTrust

Averagely improve 32.9% Averagely improve 75.6%

Page 21: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

BadRank vs. VoteTrust

Averagely improve 44.5% Averagely improve 41.6%

Page 22: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Separating Normal User from Sybils

80% with low score

Page 23: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Separating Normal User from Sybils

Maximum accuracy=85.7%

Page 24: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

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)

Page 25: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Outline

PreliminaryImplementation

– Trust-based vote assignment– Global vote aggregation

EvaluationConclusion

Page 26: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Conclusion

• VoteTrust is a rating system– high accuracy in Sybil detection– Efficient in resisting Sybil (community)

• Key techniques– Trust-based vote assignment– Global vote aggregation

Page 27: VoteTrust Leveraging Friend Invitation Graph  to  Defend Social Network Sybils

Thank you!