Welcome to Journal of University of Chinese Academy of Sciences,Today is

›› 2011, Vol. 28 ›› Issue (2): 253-261.DOI: 10.7523/j.issn.2095-6134.2011.2.018

• Research Articles • Previous Articles     Next Articles

Robust analysis of trust-based recommendation algorithms

CHEN Su, LUO Tie-Jian, XU Yan-Xiang   

  1. School of Information Science and Engineering, Graduate University, Chinese Academy of Sciences, Beijing 100049, China
  • Received:2010-03-17 Revised:2010-05-21 Online:2011-03-15
  • Supported by:

    Supported by the e-Education Project (0826011ED2) granted by the Chinese Academy of Sciences

Abstract:

Trust-based recommendation is an emerging technique,in which the trust web of users serves as an overlay to locate reliable advisers.Although this technique is claimed to be more robust than collaborative filtering in previous researches,its real strength to resist attacks has not been quantifiably studied.We propose a formal evaluation framework for this topic and compare two representative algorithms in the literature on the data set from Epinions. com.Experiments indicate the key factors for their robustness. Furthermore,several countermeasures are suggested based on these findings.

Key words: recommender system, trust metric, robustness, collaborative filtering

CLC Number: