Text Clustering

Abstract

As the saying goes, birds of a feather flock together. Clustering analysis in data mining is a process of dividing data points into different subsets based on the intrinsic rules and distribution characteristics so that the data points in the same cluster are similar to each other while those in different clusters are distinct. Each subset of data points is called a cluster. As an unsupervised machine learning method, clustering differs from classification in several ways. First, rather than requiring labeled data for supervision in the clustering process, it simply depends on a similarity computation between different data points, which therefore allows high flexibility. Second, in a classification problem, the categories should be predefined, while in clustering, the number of categories is unknown in advance. The clustering system determines the number of categories and the data points contained in each category according to certain criteria. Clustering is a fundamental problem in machine learning and has been widely used in natural language processing and text data mining.

In text clustering, the text data should first be represented in a machine computable form. Therefore, text representation is the basis of text clustering. The text representation methods have been described in Chap. 3 in detail. In this chapter, we will focus on the clustering algorithms. The typical clustering algorithms include partition-based methods, hierarchy-based methods, and density-based methods.