Accelerating Discovery

56 ◾ Accele rating Discove r y

e centroid, mentioned above, is an important concept we will be

coming back to again and again. ink of the centroid as a numeric sum-

mary of the category in dictionary space. It is a kind of model for the cat-

egory that represents what a typical document in the category looks like

in vector-space representation. What a useful thing to have; and we will

make further use of it in the very next section.

As the category editing process continues, there areas of overlap between

categories will naturally arise. As much as possible, we want to make sure

the categories are distinct, well-dened entities so that a ...