In Section 2.3, we reviewed the main classification methods: namely, nearest-
neighbors, decision trees, rule-based classifiers, Bayesian networks, artificial neural
networks, and support vector machines. We saw that, although kNN ( see Section
2.3.1) CF is the preferred approach, all those classifiers can be applied in different
settings. Decision trees ( see Section 2.3.2) can be used to derive a model based
on the content of the items or to model a particular part of the system. Decision
rules ( see Section 2.3.3) can be derived from a pre-existing decision trees, or can
also be used to introduce business or domain knowledge. Bayesian networks ( see
Section 2.3.4) are a popular approach to content-based recommendation, but can
also be used to derive a model-based CF system. In a similar way, Artificial Neu-
ral Networks can be used to derive a model-based recommender but also to com-
bine/hybridize several algorithms. Finally, support vector machines ( see Section
2.3.6) are gaining popularity also as a way to infer content-based classifications or
derive a CF model.