In previous sections we restricted our discussion of the application and use of recommender systems to the context of traditional websites. When information systems extend their reach to offer access and interaction opportunities virtually anywhere, however, the so-called ubiquitous environments become application domains for recommender systems.

In this chapter, we therefore discuss the idiosyncrasies of recommending in ubiquitous environments compared with traditional web applications. First we reflect on the evolution of mobile systems and the associated technological issues in a short introductory statement. Second, we focus on the challenges and proposed algorithms for introducing additional context data, such as location. Finally, we provide an overview of selected application domains and related work.

Introduction

Mobile applications have always been a domain for recommendation because small display sizes and space limitations naturally require access to personalized information, on one hand, and location provides an additional exploitable source of user feedback, on the other hand. Since the end of the 1990s, research into mobile applications has focused heavily on adaptivity with regards to het-erogenous hardware and software standards (see Miller et al. 2003). Therefore, most proposed mobile applications have remained in a prototypical state and have been evaluated only in small field trials with a limited scope for usage. One exception in this respect is the ClixSmart system (Smyth and Cotter 2002), which personalizes users' navigation on mobile portals and has been evaluated and fielded in real-world scenarios.