Are the recommendations from recommender system trustworthy ?
This paper has proposed a performanceoriented security solution for a personalized wellness recommender system that results in another layer of complexity making the performance requirement hard to achieve.
Abstract
The sensor data from multimodal data sources is characterized by continuous data streaming which passes through recommendation system curation layers for extensive processing to infer context-aware personalized recommendations to users. Such extensive processing of healthcare big dataset tends to introduce performance overhead during the entire life-cycle of data generation, data processing and data analysis. On the other hand, enforcing security and privacy on individual-based data is a promising requirement of healthcare datasets. However, adding security layer on such complex framework results in another layer of complexity making the performance requirement hard to achieve. Hence, a trade-off between performance and security constraints is required for wellness recommender system. In this paper we have proposed a performanceoriented security solution for a personalized wellness recommender system.