Indoor Positioning Knowledge Model for Privacy Preserving Context-Awareness

Abstract

Context-aware recommendation systems seek to provide relevant recommendation content for users by deriving preferences not just through past behavior but also based on instantaneous context information. Such responsiveness is especially crucial for offline retail setups such as supermarkets and malls, wherein user decisions map directly to how they navigate within the venue. A key area of concern for such systems is management of the tradeoff between relevance of recommendation content and privacy guarantees. In this paper, we propose a system which enables dynamic service composition of context information with recommendation content, while enabling privacy configuration. We focus on the use of indoor positioning as contextual information for indoor environments such as physical retail. Central to the proposed system is a knowledge model that integrates indoor location information with that of positioning schematics as well as relationships among locations. Specifically, we show how incorporation of the ontology model with algorithms for detection of indoor location and location semantics allows for robust configuration of not just recommendation content but also privacy policies that govern the granularity of information shared for generating context-aware recommendations. This integrated knowledge model can enable various context-based offerings in the offline or physical realm, thus bridging gap between the physical as well as digital world.