all of their preferences are satisfied, a conjunctive function like the minimum should
be used. On the other hand, if some of the preferences are unlikely to be satisfied
simultaneously, e.g. the user is interested in drama and horror films, an averaging
or disjunctive function might be more reliable. We present many examples of these
broad classes of aggregation functions in Section 22.3.
In situations where it is practical to calculate item-item similarity, content-based
filtering could also be facilitated using methods that mirror those in collaborative
filtering [2]. In this case, a user profile might consist of all or a subset of previously
rated/purchased items, D = {d1, ..., dq}, and a measure of similarity is calculated
between the unseen item di and those in D,