The data on the Web is increasingly being centralised towards a few service providers. Personal Data Stores (PDS) have emerged, proposing a fundamental shift from the current service-centric data ecosystem to a decentralised data storage and processing environment by placing the data with users. Users are to assume total self-sovereignty over their data, including opportunities to monetise. While PDS systems enable user empowerment, they also put a greater burden on the users, who may not be technically-savvy, to manage data access, which may increase the chance of unintended mishaps and privacy risks. This research proposes a privacy preference recommender system for privacy-preserving data sharing control that is designed to work with the constraints of user-centric data storage and processing environment for PDS. The outcome contributes towards a user-assisting privacy technology that utilises user context effectively to recommend privacy settings while conforming to the PDS architecture by storing and processing all analytics locally.