ABSTRACT
There is a large difference between what humans and computers perceive as a set of clusters in either the real world or in the representations of it we build with data. This chapter focuses on Human Point of View (HPOV). The main point of this chapter is that the basic ideas underlying the words similar, close, and group (compactness and separation) for humans must somehow be carried over to what we want them to mean for sets of data residing in computers. And the fact that we do disagree about what we see in the simple figures in this chapter tells us not to be very surprised if algorithms disagree about substructure in the data they process. Petrie’s statement contains all the key words of HPOV clustering: similar, groups, and closeness.
