HYBRID RECOMMENDER SYSTEMS SURVEY AND EXPERIMENTS is largely a function of the quality and quantity of data available for learning. with knowledge-based recommenders, it is a function of the available knowledge base. e can distinguish two cases: the uniform case, in which one recommender has better accuracy than another over the whole space of recommendation, and the non-uniform case, in which the two recommenders have different strengths in dif- ferent parts of the space. If the recommenders are uniformly unequal, it may make sense to employ a hybrid in which the inaccuracies of the weaker recommender can be contained: for example, a cascade scheme with the stronger recommender given higher priority, an augmentation hybrid in which the weaker recommende acts as a'bot contributing a small amount of information, or a meta-level combi- nation in which the stronger technique produces a dense representation that strengthens the performance of the weaker one. In the non-uniform case, the system hybrid is a natural choice here, but it requires that the system be able to detect li a will need to be able to employ both recommenders at different times. A switcher one recommender should be preferred Feature combination and mixed hybrids can be used to allow output from both recommenders without having to implement a switching criterion. More research is needed to establish the tradeoffs betweer these hybridization options. 4. a knowledge-based restaurant recommender system As the overview shows, there are a number of areas where the space of hybrid rec ommendation is not fully explored. In particular, there are few examples that incor- porate knowledge-based recommendation. Knowledge-based recommendation is at the heart of a research program known as Find-Me Systems'(Burke et al., 1997; Burke 1999; 2000). The restaurant recommender Entree is one example of such a system. This section provides a brief overview of Entree, and then introduces EntreeC, a hybrid recommender system that adds collaborative filtering to Entree, creating a knowledge-based/collaborative cascade hybrid 4. 1. ENTREE Entree is a restaurant recommendation system that uses case-based reasoning Kolodner, 1993)techniques to select and rank restaurants. It was implemented to serve as a guide to attendees of the 1996 Democratic National Convention in Chicago and has been operating as a web utility since that time A user interacts with the system by submitting an entry point, either a known restaurant or a set of criteria, and is shown similar restaurants. The user then interacts with the system in a dialog, critiquing the systems suggestions and interactively refining the search until an acceptable option is achieved. Consider a user who starts browsing by entering a query in the form of a known restaurant, Wolfgang Puck's'Chinois on Main'in Los Angeles, as shown in Figure 1.(The Url:http://infolab.ilsnwu.edu/entree Reproduced with permission of the copyright owner. Further reproduction prohibited without permissionReproduced with permission of the copyright owner. Further reproduction prohibited without permission