Special issue on Recommender Systems

More than fifteen years after the creation of the first software systems now known as Recommender Systems (RS), the research area is more active than ever. The inception of an ACM-supported conference series in 2007, the series of yearly workshops concerned with RS organized in conjunction with international conferences, as well as recently published special issues in highly visible journals and magazines all document this fact. The reasons for this increased interest in Recommender Systems, i.e. systems that help online users deal with the information overload and support them in decision-making and purchase processes, are manifold. On the one hand, the Web of today is now well-established as a trusted medium not only for information gathering but also for actually purchasing products or services. Correspondingly, there are more online customers, more online shops and more products sold over the web than ever before. In this context, Recommender Systems are not only able to help more users to separate the information they need from a vast number of offerings, but also represent a valuable opportunity for shop owners to differentiate themselves from their competitors by offering an online service that is augmented with a recommendation facility. On the other hand, the way the web is used has also changed dramatically. Until recently, information on the web was primarily static and sites were established and maintained by their respective owners. Today, in the age of Web 2.0, users not only actively and willingly share information with other users but the available pieces of information are also highly interrelated with, e.g. in the form of social networks. These two aspects are driving research in the area as user feedback and user participation are particularly important for the success of Recommender Systems in general. In addition, today’s and tomorrow’s Recommender Systems will increasingly exploit information that goes beyond simple user and item ratings for generating recommendations, utilizing more advanced concepts, for instance, the trust relationships present in a social network. This special issue contains a collection of original scientific papers that document the latest advances in Recommender Systems research and gives an overview of current topics in the field. The selection process for papers contained in this special issue was very restrictive and subsequently only four regular papers and four shorter research notes have been accepted from a total of 38 submissions. The regular papers are covering the following topics. On Exploiting Taxonomies for Recommender Systems by Ziegler et al. shows how existing taxonomic information derived from catalog items can be used to improve the prediction quality of a collaborative recommender system. This paper follows the recent trend of exploiting additional information (concerning the item’s “content” or the users “context”) in recommender systems hybrids. In addition, the paper contains an overview of classical technology and – as another recent topic – discusses aspects of diversity in recommendation lists.