Deep learning personalized recommendation-based construction method of hybrid blockchain model

This study aims to explore the construction of a personalized recommendation system (PRS) based on deep learning under the hybrid blockchain model to further improve the performance of the PRS. Blockchain technology is introduced and further improved to address security problems such as information...

Full description

Saved in:
Bibliographic Details
Published in:Scientific reports 2023-10, Vol.13 (1), p.17915-17915, Article 17915
Main Authors: Yu, Xiaomo, Li, Wenjing, Zhou, Xiaomeng, Tang, Ling, Sharma, Rohit
Format: Article
Language:English
Subjects:
631/1647
639/4077
Algorithms
Blockchain
Deep learning
Fault tolerance
Humanities and Social Sciences
multidisciplinary
Network security
Science
Science (multidisciplinary)
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This study aims to explore the construction of a personalized recommendation system (PRS) based on deep learning under the hybrid blockchain model to further improve the performance of the PRS. Blockchain technology is introduced and further improved to address security problems such as information leakage in PRS. A Delegated Proof of Stake-Byzantine Algorand-Directed Acyclic Graph consensus algorithm, namely PBDAG consensus algorithm, is designed for public chains. Finally, a personalized recommendation model based on the hybrid blockchain PBDAG consensus algorithm combined with an optimized back propagation algorithm is constructed. Through simulation, the performance of this model is compared with practical Byzantine Fault Tolerance, Byzantine Fault Tolerance, Hybrid Parallel Byzantine Fault Tolerance, Redundant Byzantine Fault Tolerance, and Delegated Byzantine Fault Tolerance. The results show that the model algorithm adopted here has a lower average delay time, a data message delivery rate that is stable at 80%, a data message leakage rate that is stable at about 10%, and a system classification prediction error that does not exceed 10%. Therefore, the constructed model not only ensures low delay performance but also has high network security performance, enabling more efficient and accurate interaction of information. This solution provides an experimental basis for the information security and development trend of different types of data PRSs in various fields.
ISSN:2045-2322
2045-2322
DOI:10.1038/s41598-023-39564-x