PROPOSING A MODEL TO GENERATE RECOMMENDATION TO E-BUYERS BASED ON TRUST EVALUATION

PROPOSING A MODEL TO GENERATE RECOMMENDATION TO E-BUYERS BASED ON TRUST EVALUATION

Arman Zaheri1 Mohammad Fathian2 Seyyed-Mahdi Hosseini-Motlagh3 Masoud Mansoury4

1) M.Sc. Degree in Information Technology, Iran University of Science and Technology, Tehran, Iran -
2) Professor and Ph.D. Degree in Industrial Engineering, Iran University of Science and Technology, Tehran, Iran
3) Ph.D. Degree in Industrial Engineering, Iran University of Science and Technology, Tehran, Iran
4) M.Sc. Degree in Information Technology, Amir Kabir, Tehran, Iran -

IOI Code : XYGF-BDCAD
Publication : 3th International Conference on management and industrial engineering(3icmi.com)
Abstract :
    Recommender systems using traditional collaborative filtering suffer from some significant weaknesses, such as cold-start users and data sparsity. Due to data sparsity of the user-item matrix, we may not find the similar neighbors of the active user and thus collaborative filtering Recommender Systems often fails in predicting items ratings. Considering trust can alleviate those problems. In such systems, users’ explicit trust rating can be used for neighbor formation to generate automated recommendation. However, reliable explicit trust data is not always available. In this paper we propose a new method of deriving implicit trust relationships based on user’s interest similarity without needing explicit trust data. We consider replacing similarity weight with implicit trust weight over the trust network. To measure the users interest similarity, user’s personalized tagging information is used. The result of evaluating our approach on Epinions dataset shows that this approach can improve accuracy and coverage of recommender systems.

Keywords : recommender system implicit trust similarity items ratings