An intelligent client-centric framework for responsive, privacy conscious personalisation

Abstract

Personalisation is used extensively to improve user engagement, to optimise user experience and to enhance marketing and advertising online. However, it inherently requires the collection of users' personal information. In recent years, privacy has become a core consideration of consumers, providers, and governments. This is due to rising unauthorised data breaches as well as increasing awareness of the value and private nature of data. Many now market their products and services based on privacy and Governments have also recently begun to take action, for example, through GDPR. Users are becoming less likely to share or give their personal information to private companies, who are then limited in the level of personalisation they can provide. This has resulted in a personalisation-privacy paradox. The more information there is about a user, the better the system can adapt to the user's needs, but the less the privacy of the user's personal data is protected.

Privacy-conscious web frameworks such as Client-Side Personalisation (CSP), attempt to shift the data storage approach, storing user data with trusted third parties or on the client's own device. This keeps the user data under the control of the client, allowing users to enjoy personalised content without compromising the privacy of their personal data. However, this comes with a number of issues such as intellectual property concerns,limited inference data, infeasible data overheads, restrictive client processing power and the inability to easily sync profiles across devices. Architectures have been proposed which tackle most of these issues, however, these solutions still have signifcant problems with scalability and performance.

This thesis introduces the Intelligent Client-Centric Personalisation (ICCP) framework. This builds on existing frameworks augmenting them with predictive prefetching to tackle performance issues and reduce user latency. This responsive, lightweight, privacyconscious approach to personalisation enables companies to continue offering personalisation services while ensuring that consumers hold onto and have control over their personal data.

The research detailed in the thesis examines the design challenges, privacy and performance trade-offs of an ICCP framework through extensive examination and discussion. The thesis evaluates the framework, focusing on the effect of the prefetching strategy on user latency as well as the effect of the framework configurations on this latency. Ultimately, the thesis outlines the circumstances under which an ICCP framework provides marked benefits over alternative frameworks.