Modeling Trust for Recommender Systems using Similarity Metrics 117 We coded our derived trust opinions into metrics taken from Shaferian belief theory and we attempted an evaluation of our model by comparing the resulting si- milarity to that derived from secondary trust. We also compared the proposed approach against another that has been used in the past and in which the shaping of the derived uncertainty is dependent on a pre dictability measure and thus on the quality of the evidence. The comparison showed that using qualitative measures for deriving trust not only incur a computation pe nalty but also provide lower accuracy when compared with less complex approaches for describing the users behavior In conclusion, the strong points of the proposed technique can be summarized as its ability to incorporate similarity measures in its properties, the low computation complexity and its flexibility in accepting datasets in which user ratings are ex pressed in continuous values. In terms of accuracy in deriving trust opinions,a comparison against the older al ternative shows that the new one is more than twice as accurate. We would suggest that the method is very suitable for use in CF re commander systems References I.http://www.epinions.com 2.http://www.ebay.com 3.http://www.am mazon. com 4. D. Goldberg, D. Nichols, B M. Oki, D.Terry, Using Collaborative filtering to weave an info mation tapestry", Communication of ACM, 35(12): 61-70, 1992 5. B Sarwar, G Karypis, J Konstan, J Reidl,"Analysis of Recommendation Algorithms for ECommerce, In Proceedings of the Second ACM Conference on Electronic Commerce pg 158-168, ACM Press2000. 6. M P. OMahony, N.J. Hurley, G.C. M. Silvestre, " Detecting noise in recommender system da tabases", Proceedings of the l lth international conference on Intelligent user interfaces, Janu- ary 29-February 01, 2006, Sydney, Australia 7. B. MSarwar, J.T. Riedl, "Sparsity, scalability, and distribution in recommender systems, 2001, ISBN: 0-493-04207-5, University of Minnesota. 8. A Josang, E. Gray and M. Kinateder. Analysing Topologies of Transitive Trust", In the pro- ceedings of the Workshop of Formal Aspects of Security and Trust(FAST) 2003, Pisa, Sep- tember 2003 9. G Pitsilis. L F. Marshall. "A model for trust derivation from evidence for use in recommende systems", Newcastle University, Technical report CS-TR-874, 2004 10. AJosang,"A Logic for Uncertain probabilities", International Journal of Uncertainty, fuzzi- ness and Knowledge based systems, Vol 9, No 3, June 200 I. S. Marsh, "Formalizing Trust as Computational concept", PhD Thesis, University of Stirling, cotland 1994 12. A Rahman, S Heiles, Supporting trust in Virtual Communities", In proceedings of Interna- tional conference On System Sciences, Jan 4-7-2000, Hawaii 13. G. Shafer."A Mathematical Th f Evidence". Princeton University Press. 1976 14. A Josang, R Ismail. "The Beta Reputation System". In the proceedings of the 15th Bled Conference on Electronic Commerce, Bled, Slovenia, 17-19 June 2002 15. CC Aggarwal, J. L Wolf, K Wu, P.S.Yu,. " Horting Hatches an Egg: A New Graph-theoretic Approach to Collaborative Filtering, In Proceedings of the ACM KDD99 Conference. San 16. P. Massa-P. Avesani,Trust-aware Collaborative Filtering for recommender Systems", Coo- pIS/DOA/ODBASE (1)2004: 492-508We coded our derived trust opinions into metrics taken from Shaferian belief theory and we attempted an evaluation of our model by comparing the resulting si￾milarity to that derived from secondary trust. We also compared the proposed approach against another that has been used in the past and in which the shaping of the derived uncertainty is dependent on a pre￾dictability measure and thus on the quality of the evidence. The comparison showed that using qualitative measures for deriving trust not only incur a computation pe￾nalty but also provide lower accuracy when compared with less complex approaches for describing the user’s behavior. In conclusion, the strong points of the proposed technique can be summarized as its ability to incorporate similarity measures in its properties, the low computation complexity and its flexibility in accepting datasets in which user ratings are ex￾pressed in continuous values. In terms of accuracy in deriving trust opinions, a comparison against the older alternative shows that the new one is more than twice as accurate. We would suggest that the method is very suitable for use in CF re￾commender systems. References 1. http://www.epinions.com 2. http://www.ebay.com 3. http://www.amazon.com 4. D.Goldberg, D.Nichols, B.M.Oki, D.Terry, “Using Collaborative filtering to weave an infor￾mation tapestry”, Communication of ACM, 35(12):61-70, 1992 5. B.Sarwar, G.Karypis, J.Konstan, J.Reidl, “Analysis of Recommendation Algorithms for ECommerce”, In Proceedings of the Second ACM Conference on Electronic Commerce pg 158-168,ACM Press 2000. 6. M.P. O'Mahony , N.J. Hurley , G.C.M. Silvestre, “Detecting noise in recommender system da￾tabases”, Proceedings of the 11th international conference on Intelligent user interfaces, Janu￾ary 29-February 01, 2006, Sydney, Australia 7. B. M.Sarwar , J.T.Riedl, “Sparsity, scalability, and distribution in recommender systems,” 2001, ISBN:0-493-04207-5, University of Minnesota. 8. A. Jøsang, E. Gray and M. Kinateder. “Analysing Topologies of Transitive Trust”, In the pro￾ceedings of the Workshop of Formal Aspects of Security and Trust (FAST) 2003, Pisa, Sep￾tember 2003. 9. G.Pitsilis., L.F.Marshall, “A model for trust derivation from evidence for use in recommender systems”, Newcastle University, Technical report CS-TR-874, 2004. 10. A.Josang, “A Logic for Uncertain probabilities”, International Journal of Uncertainty, fuzzi￾ness and Knowledge based systems, Vol.9,No.3, June 2001. 11. S.Marsh, “Formalizing Trust as Computational concept”, PhD Thesis, University of Stirling, Scotland 1994. 12. A.Rahman, S.Heiles,”Supporting trust in Virtual Communities”, In proceedings of Interna￾tional conference On System Sciences, Jan 4-7-2000, Hawaii. 13. G.Shafer, “A Mathematical Theory of Evidence”, Princeton University Press. 1976 14. A. Jøsang, R. Ismail. “The Beta Reputation System”. In the proceedings of the 15th Bled Conference on Electronic Commerce, Bled, Slovenia, 17-19 June 2002. 15. C.C.Aggarwal, J.L.Wolf, K.Wu, P.S.Yu,.”Horting Hatches an Egg: A New Graph-theoretic Approach to Collaborative Filtering”, In Proceedings of the ACM KDD’99 Conference. San Diego, CA, pp.201-212. 16. P.Massa – P.Avesani, ”Trust-aware Collaborative Filtering for recommender Systems”, Coo￾pIS/DOA/ODBASE (1) 2004: 492-508 Modeling Trust for Recommender Systems using Similarity Metrics 117