TrustWalker: A Random Walk Model for Combining Trust-based and Item-based Recommendation

1
TrustWalker A Random Walk Model for Combining
Trust-based and Item-based Recommendation

• Mohsen Jamali Martin Ester
• Simon Fraser University, Vancouver, Canada

2
Outline

• Introduction
• TrustWalker
• Single Random Walk
• Recommendation
• Matrix Notation
• Properties of TrustWalker
• Confidence, Special Extreme Cases
• Experiments
• Conclusion and Future Work

3
Introduction - Recommendation

• Need For Recommenders
• Problem Definition
• Given user u and target item i
• Predict the rating ru,i
• Collaborative Filtering
• Considers Users with Similar Rating Patterns
• Aggregates the ratings of Similar Users

4
Introduction Trust-based RS

• Issues with CF
• Requires Enough Ratings (Cold Start Users)
• Vulnerable to Attack Profiles
• Social Networks Emerged Recently
• Independent source of information
• Motivations of Trust-based RS
• Social Influence users adopt the behavior of
  their friends

5
Trust-based Recommendation

• Explores the trust network to find Raters.
• Aggregate the ratings from raters for prediction.
• Different weights for users
• 510818

• Advantages
• Improving the coverage
• Attack resistance

6
TrustWalker - Motivation

• Issues in Trust-based Recommendation
• Noisy data in far distances
• Low probability of Finding rater at close
  distances

7
TrustWalker - Motivation

• How Far to Go into Network?
• Tradeoff between Precision and Recall
• Trusted friends on similar items
• Far neighbors on the exact target item

8
TrustWalker

• TrustWalker
• Random Walk Model
• Combines Item-based Recommendation and
  Trust-based Recommendation
• Random Walk
• To find a rating on the exact target item or a
  similar item
• Prediction returned rating

9
Single Random Walk

• Starts from Source user u0.
• At step k, at node u
• If u has rated I, return ru,i
• With Fu,i,k , the random walk stops
• Randomly select item j rated by u and return ru,j
  .
• With 1- Fu,i,k , continue the random walk to a
  direct neighbor of u.

10
Item Similarities in TrustWalker

• Item Similarities
• Probability of having high correlation for pairs
  of items with few users in common is high.

11
Stopping Probability in TrustWalker

• Fu,i,k
• Similarity of items rated by u and target item i.
• The step of random walk

12
Recommendation in TrustWalker

• Prediction Expected value of rating returned by
  random walk.

13
Performing Random Walks

• Matrix Notation for TrustWalker
• Expensive
• We perform actual random walks
• Result of a Single Random Walk is not precise
• We perform several random walks
• Prediction Average of results
• The variance of results of different random walk
  converges

14
Properties of TrustWalker

• Special Cases of TrustWalker
• Fu,i,k 1
• Random Walk Never Starts.
• Item-based Recommendation.
• Fu,i,k 0
• Pure Trust-based Recommendation.
• Continues until finding the exact target item.
• Aggregates the ratings weighted by probability of
  reaching them.
• Existing methods approximate this 510.
• Confidence
• How confident is the prediction

15
Related Work

• Tidal Trust 5
• BFS to find raters at the closest distance
• Mole Trust 10
• BFS to find rater up to depth max-depth
• aggregate the ratings according to the trust
  values of the rater and the source user
• Item-based CF 15
• Aggregate the ratings of source users on similar
  items weighted by their similarities.

16
Experiments

• Epinions.com Data Set
• 49K users, 24K cold start users ( users with less
  than 5 ratings)
• 104K items, 575K ratings, 508K trust expressions
• Binary trust, ratings in 1,5
• Leave-one-out method
• Evaluation Metrics
• RMSE
• Coverage
• Precision 1- RMSE/4

17
Comparison Partner

• Tidal Trust 5
• Mole Trust 10
• CF Pearson
• Random Walk 6,1
• Item-based CF
• TrustWalker0 -pure
• TrustWalker -pure

18
Experiments Cold Start Users
19
Experiment- All users
20
Experiments - Confidence

• More confident Predictions have lower error

21
Conclusion

• Conclusion
• Random Walk Method
• Combines Trust-based and Item-based
  Recommendation.
• Computes the confidence in Predictions
• Includes existing recommenders in its special
  cases.
• Future Directions
• Top-N recommendation RecSys09
• Distributed Recommender
• Context dependent trust

22
Thank You
23
References

• 1 R. Andersen, C. Borgs, J. Chayes, U. Feige,
  A. Flaxman, A. Kalai, V. Mirrokni, and M.
  Tennenholtz. Trust-based Recommendation systems
  an axiomatic approach. In WWW 2008.
• 2 R. M. Bell, Y. Koren, and C. Volinsky.
  Modeling relationships at multiple scales to
  improve accuracy of large recommender systems. In
  KDD 2007.
• 3 S. Brin and L. Page. The anatomy of a
  large-scale hypertextual web search engine.
  Computer Networks and ISDN Systems, 30(1), 1998.
• 4 D. Crandall, D. Cosley, D. Huttenlocher, J.
  Kleinberg, and S. Suri. Feedback effects between
  similarity and social influence in online
  communities. In KDD 2008.
• 5 J. Golbeck. Computing and Applying Trust in
  Web-based Social Networks. PhD thesis, University
  of Maryland College Park, 2005.
• 6 D. Goldberg, D. Nichols, B. M. Oki, and D.
  Terry. Using collaborative ltering to weave an
  information tapestry. Communications of the ACM,
  35(12), 1992.

24
References

• 7 Y. Koren. Factorization meets the
  neighborhood a multifaceted collaborative
  ltering model. In KDD 2008.
• 8 Levien and Aiken. Advogato's trust metric.
  online at http//advogato.org/trust-metric.html,
  2002.
• 9 H. Ma, H. Yang, M. R. Lyu, and I. King.
  Sorec social recommendation using probabilistic
  matrix factorization. In CIKM '08, 2008.
• 10 P. Massa and P. Avesani. Trust-aware
  recommender systems. In ACM Recommender Systems
  Conference (RecSys), USA, 2007.
• 11 S. Milgram. The small world problem.
  Psychology Today, 2, 1967.
• 12 J. O'Donovan and B. Smyth. Trust in
  recommender systems. In 10th international
  conference on Intelligent user interfaces, USA,
  2005.

25
References

• 13 A. Rettinger, M. Nickles, and V. Tresp. A
  statistical relational model for trust learning.
  In AAMAS '08 7th international joint conference
  on Autonomous agents and multiagent systems,
  2008.
• 14 M. Richardson and P. Domingos. Mining
  knowledge-sharing sites for viral marketing. In
  KDD 2002.
• 15 B. Sarwar, G. Karypis, J. Konstan, and J.
  Riedl. Item-based collaborative filtering
  recommendation algorithms. In WWW 2001.
• 16 S. Wasserman and K. Faust. Social Network
  Analysis. Cambridge Univ. Press, 1994.
• 17 H. Yildirim and M. S. Krishnamoorthy. A
  random walk method for alleviating the sparsity
  problem in collaborative filtering. In ACM
  Conference on Recommender Systems (RecSys),
  Switzerland, 2008.
• 18 C. N. Ziegler. Towards Decentralized
  Recommender Systems. PhD thesis, University of
  Freiburg, 2005.

26
TrustWalker
?
27
TrustWalker
5
28
TrustWalker
4
R1 5
Continue?
Yes
?
29
TrustWalker
R1 5 R2 4
Continue?
Yes
Continue?
R3 5
Yes
Continue?
No
5
Prediction 4.67