Case-Based Recommender Systems

In the past years, a number of research projects have focused on recommender systems. These systems implement various learning strategies to collect and induce user preferences over time and automatically suggest products that fit the learned user model. The most popular recommendation methodology is collaborative filtering (Resnick, Iacovou, Suchak, Bergstrom, & Riedl, 1994) that aggregates data about customer’s preferences (ratings) to recommend new products to the customers. Content-based filtering (Burke, 2000) is another approach that builds a model of user interests, one for each user, by analyzing the specific customer behavior. In collaborative filtering the recommendation depends on the previous customers’ information, and a large number of previous user/system interactions are required to build reliable recommendations. In content-based systems only the data of the current user are exploited and it requires either explicit information about user interest, or a record of implicit feedback to build a model of user interests. Content-based systems are usually implemented as classifier systems based on machine learning research (Witten & Frank, 2000). In general, both approaches do not exploit specific knowledge of the domain. For instance, if the domain is computer recommendation, the two above approaches, in building the recommendation for a specific customer, will not exploit knowledge about how a computer works and what is the function of a computer component. Conversely, in a third approach called knowledgebased, specific domain knowledge is used to reason about what products fit the customer’s preferences (Burke, 2000). The most important advantage is that knowledge can be expressed as a detailed user model, a model of the selection process or a description of the items that will be suggested. Knowledge-based recommenders can exploit the knowledge contained in case or encoded in a similarity metric. Case-Based Reasoning (CBR) is one of the methodologies used in the knowledge-based approach. CBR is a problem solving methodology that faces a new problem by first retrieving a past, already solved similar case, and then reusing that case for solving the current problem (Aaamodt & Plaza, 1994). In a CBR recommender system (CBR-RS) a set of suggested products is retrieved from the case base by searching for cases similar to a case described by the user (Burke, 2000). In the simplest application of CBR to recommendation problem solving, the user is supposed to look for some product to purchase. He/she inputs some requirements about the product and the system searches in the case base for similar products (by means of a similarity metric) that match the user requirements. A set of cases is retrieved from the case base and these cases can be recommender to the user. If the user is not satisfied with the recommendation he/she can modify the requirements, i.e. build another query, and a new cycle of the recommendation process is started. In a CBR-RS the effectiveness of the recommendation is based on: the ability to match user preferences with product description; the tools used to explain the match and to enforce the validity of the suggestion; the function provided for navigating the information space. CBR can support the recommendation process in a number of ways. In the simplest approach the CBR retrieval is called taking in input a partial case defined by a set of user preferences (attribute-value pairs) and a set of products matching these preferences are returned to the user.