Can a classifier be used to measure the Clustering Quality measure?
I have come across this paper where the researcher uses a classifier like : five nearest neighbor classifier, a C4.5 decision tree, and a Rocchio classifier .
But I'm not sure how these classifiers help in the Measuring the clustering quality of a dataset. Can anyone explain how the different classifiers help in the Measuring the clustering quality of a dataset?
The paper says in section 3 that "The basis of our new CQM is the notion
that clustering’s purpose is the unsupervised prediction of
behavior from populations. While prediction suggests some
form of classification, there are typically no labels during
clustering, preventing classification’s use. However, in the
context of a CQM, which is computed after a clustering
is complete, we do in-fact have a labeling that a classifier
can use—the clustering itself."
Could not understand
https://bitbucket.org/parashararamsees/clustering-quality-measure/downloads