SCIENCE@home

ADVANCING CROSS-DOMAIN RECOMMENDATION FOR SECURE RATING PREDICTION THROUGH FEDERATED LEARNING: A SYSTEMATIC REVIEW

P. Patil,Anuradha Kanade

Published 2025 in International Journal of Applied Mathematics

ABSTRACT

Aim: This study evaluates the efficacy of federated cross-domain system recommendations in comparison to conventional recommendation systems for privacy-preserving rating prediction. Through the investigation of the interaction between recommendation systems and federated learning, this research can reduce the privacy issue, data security, and the barriers brought by integration of cross-domain data, and further evaluate the potential for more personalized and secure recommendation systems. Methodology: A thorough literature review with focus on employing multi-domain recommendation systems to federated learning was conducted. Search on multiple academic platforms provided relevant data and publications on this subject matter. The selected studies and datasets were reviewed based on set inclusion and exclusion criteria to enable comparison with classical recommendation systems. Data was then compiled and used to evaluate the limitations, privacy-preservation methods, and performance of federated recommendation systems. Results: Following an initial search of 1188 records, 71 studies were deemed eligible for publication. Relative to traditional methods, federated learning does hold great promise for enhanced security and privacy in recommendation systems. On data heterogeneity, scalability, and model performance across domains, the picture was not as bright. Among the key issues raised were data integration across domains, algorithmic bias, and the need for more robust privacy-preserving technology. Findings: The work illustrates that federated cross-domain recommendation systems may effectively reconcile privacy protection with recommendation accuracy, particularly when using privacy-preserving techniques like differential privacy and safe multi-party computing. Although federated systems have more robust privacy protection than conventional recommendation methods, they have more challenges in data sparsity and domain adaption. Conclusion: This paper offers insightful analysis of how federated learning can be utilized to produce cross-domain recommendation systems with the maintenance of privacy. For the sake of ensuring the universal acceptance and effectiveness of federated recommendation systems, especially compared to traditional models, future studies must focus on enhancing privacy-maintaining methods, boosting cross-domain fusion, and expanding empirical verification into real-world use. should concentrate on improving privacy-preserving techniques, enhancing cross-domain integration, and extending empirical validation to real-world applications.  

PUBLICATION RECORD

CITATION MAP

EXTRACTION MAP

CLAIMS

  • No claims are published for this paper.

CONCEPTS

  • No concepts are published for this paper.

REFERENCES

Showing 1-48 of 48 references · Page 1 of 1

CITED BY

  • No citing papers are available for this paper.

Showing 0-0 of 0 citing papers · Page 1 of 1