We use cookies to improve your experience with our site.
Li-e Wang, Dong-cheng Li, Peng Liu, Xian-xian Li. BAM_CRS: Blockchain-based Anonymous Model for Cross-domain Recommendation Systems[J]. Journal of Computer Science and Technology. DOI: 10.1007/s11390-021-0657-9
Citation: Li-e Wang, Dong-cheng Li, Peng Liu, Xian-xian Li. BAM_CRS: Blockchain-based Anonymous Model for Cross-domain Recommendation Systems[J]. Journal of Computer Science and Technology. DOI: 10.1007/s11390-021-0657-9

BAM_CRS: Blockchain-based Anonymous Model for Cross-domain Recommendation Systems

  • The recommendation system is a matter of great importance for modern e-commerce platforms to obtain more customer resources. With the quantities of users and products growing, the cold start problem is becoming further aggravated. The cross-domain recommendation technique has been widely applied in recommendation systems, which can solve cold start problems effectively. However, cross-domain recommendation introduces additional data security and authenticity challenges due to the different sources of data, which can bring more information for recommendations and more background knowledge for adversaries. And some dishonest participants may provide false information for their own benefits or selfish reasons. In this work, we propose a blockchain-based anonymous model for cross-domain recommendation systems (BAM_CRS) for guaranteeing the data authenticity and security. Under this model, we use a three-chain structure to store users, products and the relationships between the two parties, which enables the separation of the transaction and users to establish consensus without relying on a central authority. On the one hand, the recommender center only collects the relationship chain with no information of users or products to guarantee privacy. On the other hand, we design a signature-based verifiable mechanism to prevent the data from being tampered with for ensuring the accuracy of the recommendation based on unmodified data. Specifically, we also design a contribution-based quantification and incentive mechanism to motivate the participation of nodes and guarantee data authenticity. Experiments were performed on real-life datasets to evaluate the approach’s performance by comparing with competitors’ works. The results of experiment demonstrate that our model can protect data security and improve the precision of a recommendation system effectively.
  • loading

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return