1 computing trust in social networks jennifer golbeck college of information studies

34
Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

Post on 20-Dec-2015

214 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

1

Computing Trust in Social Networks

Jennifer GolbeckCollege of Information Studies

Page 2: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

2

Web-Based Social Networks (WBSNs)

• Websites and interfaces that let people maintain browsable lists of friends

• Last count– 245 social networking websites– Over 850,000,000 accounts– Full list at http://trust.mindswap.org

Page 3: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

3

Using WBSNs

• Lots of users, spending lots of time creating public information about their preferences

• We should be able to use that to build better applications

• When I want a recommendation, who do I ask?– The people I trust

Page 4: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

4

Applications of Trust

• With direct knowledge or a recommendation about how much to trust people, this value can be used as a filter in many applications

• Since social networks are so prominent on the web, it is a public, accessible data source for determining the quality of annotations and information

Page 5: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

5

Research Areas

• Inferring Trust Relationships• Using Trust in Applications

Page 6: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

6

Inferring Trust

The Goal: Select two individuals - the source (node A) and sink (node C) - and recommend to the source how much to trust the sink.

A B CtAB tBC

tAC

Page 7: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

7

Methods

• TidalTrust– Personalized trust inference algorithm

• SUNNY– Bayes Network algorithm that

computes trust inferences and a confidence interval on the inferred value.

• Profile Based– Trust from similarity

Page 8: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

8

SourceSink

Page 9: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

9

Trust Algorithm

• If the source does not know the sink, the source asks all of its friends how much to trust the sink, and computes a trust value by a weighted average

• Neighbors repeat the process if they do not have a direct rating for the sink

Page 10: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

10

Accuracy

• Comparison to other algorithms – Beth-Borcherding-Klein (BBK) 1994

  AlgorithmNetwork TidalTrust BBKTrust Project 1.09 (.99) 1.59 (1.42)FilmTrust 1.35 (1.23) 2.75 (1.80)

Page 11: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

11

Trust from Similarity

• We know trust correlates with overall similarity (Ziegler and Golbeck, 2006)

• Does trust capture more than just overall agreement?

• Two Part Analysis– Controlled study to find profile similarity measures

that relate to trust– Verification through application in a live system

Page 12: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

12

Experimental Outline

• Phase 1: Rate Movies - Subjects rate movies on the list– Ratings grouped as extreme (1,2,9,10) or far

from average (≥4 different)

• Create profiles of hypothetical users – Profile is a list of movies and the

hypothetical user’s ratings of them

• Subjects rate how much they would trust the person represented by the profile– Vary the profile’s ratings in a controlled way

Page 13: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

13

Generating Profiles

• Each profile contained exactly 10 movies, 4 from an experimental category and 6 from its complement– E.g. 4 movies with extreme ratings and 6

with non-extreme ratings

• Control for average difference, standard deviation, etc. so we could see how differences on specific categories of films affected trust

Page 14: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

14

Example Profile

• Movies m1 through m10

• User ratings r1…r10 for m1…m10

– r1…r4 are extreme (1,2,9, or 10)

– r5…r10 are not extreme

• Profile ratings pi = ri§i

Page 15: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

15

Results

1. Reconfirmed that trust strongly correlates with overall similarity ().

2. Agreement on extremes ()

3. Largest single difference (r)4. Subject’s propensity to trust ()

Page 16: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

16

• When high are used on movies with extreme ratings, the trust ratings are significantly lower than when low are applied to those films

• Statistically significant for all i

Extreme Ratings

Page 17: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

17

Maximum Difference (r)

• Holding overall agreement and standard deviation constant, trust decreased as the single largest difference between the profile and the subject (r) increased.

Page 18: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

18

Propensity to Trust ()

Page 19: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

19

Validation• Gather all pairs of FilmTrust users who

have a known trust relationship and share movies in common – 322 total user pairs

• Develop a formula using the experimental parameters to estimate trust

• Compute accuracy by comparing computed trust value with known value

Page 20: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

20

In FilmTrust

Use weights (w1,w2, w3, w4, w) = (7,2,1,8,2)

Overall Similarity

Only

Our Formula

Correlation 0.24 0.73Absolute Mean Error 1.91 1.13Std. Dev of Mean Error 1.95 0.95

Page 21: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

21

Effect of change

• If a node changes it’s trust value for another, that will propagate through the inferred values

• How far? What is the magnitude? Does the impact increase or decrease with distance?

• How does this relate to the algorithm?

• Joint work with Ugur Kuter

Page 22: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

22

Page 23: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

23

Algorithms Considered

• Eigentrust– Global algorithm– Like PageRank, but with weighted edges

• Advogato– Finds paths through the network– Global group trust metric that uses a set of

authoritative nodes to decide how trustworthy a person is

• TidalTrust• TidalTrust++

– No minimum distance - search the entire network

Page 24: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

24

Initial ideas?

• The further you get from the sink, the smaller the impact.

• Changes by more central, highly connected nodes will create a bigger impact.

Page 25: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

25

Network

Page 26: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

26

Methodology

• Pick a pair of nodes in the network– Set trust to 0– Infer trust values between all pairs– Set trust to 1– Infer trust value between all pairs– Compare inferred values from trust=0 to

trust=1

• Repeat for every pair• Repeat for each algorithm

Page 27: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

27

Fraction of Nodes at a Given Distance Whose Inferred Trust Value for the Sink

Changed

Page 28: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

28

SourceSink

Page 29: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

29

Average Magnitude of Change at a Given Distance

Page 30: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

30

Conclusions andFuture Directions

Page 31: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

31

Conclusions

• Trust is an important relationship in social networks.

• Social relationships are different than other common data used in CS research.

• Trust can be computed in a variety of ways

• The type of algorithm and behavior of users in the network impact the stability of trust inferences

Page 32: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

32

Future Work - Computing with Trust

• Major categories of trust inference: global vs. local, same scale vs. new scale– All have algorithms

• Additional features (like confidence)• Hybrid approaches

– Use trust assigned by users and similarity– Use multiple relationships for better

certainty in certain domains (e.g. authority)

Page 33: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

33

Future Work - Applications

• What sort of applications can trust be used to support?

• Recommender systems, email filtering, tagging, information ranking

• Disaster response– Highlight relevant items among vast

collections of data

Page 34: 1 Computing Trust in Social Networks Jennifer Golbeck College of Information Studies

34

• Jennifer Golbeck• [email protected]• http://www.cs.umd.edu/~golbeck