Most clustering algorithms are unsupervised, simply, taking groups of values and associating items based on similarity of attributes. Of cousre, the idea of what similarity means for a particular dataset does have to be specified for the algorithm used, or is a fixed feature of the algorithm such as a geeral notion of a compact cluster.
Examples of unsupervised clustering algorithms include k-means and self-organising maps.
Used in Chap. 18: page 286