Analysis of Text Patterns Using Kernel Methods - Text Mining: Classification, Clustering, and Applications - page 1

Chapter 1

Analysis of Text Patterns Using
Kernel Methods

Marco Turchi, Alessia Mammone, and Nello Cristianini

1.1 Introduction ............................................................. 1

1.2 General Overview on Kernel Methods .................................. 1

1.3 Kernels for Text ......................................................... 11

1.4 Example ................................................................. 19

........................................

1.5 Conclusion and Further Reading 22

1.1 Introduction
The kernel approach offers a very general framework for performing pattern
analysis on many types of data and it can be used in a wide variety of tasks and
application areas. The kernel technique also enables us to use feature spaces
whose dimensionality is more than polynomial in the relevant parameters
of the systems even though the computational cost of the pattern analysis
algorithm remains polynomial.
Our aim in this chapter is to illustrate the key ideas underlying the pro-
posed approach, by giving a theoretical background on kernels, their main
characteristics and how to build them. Then, starting from the representa-
tion of a document, we will outline kernel functions that can be used in text
analysis.

1.2 General Overview on Kernel Methods
Throughout the chapter we will assume that we have a set of data and
that we want to detect interesting relationships within it. Text documents
will be our running example. In the following pages we will not assume any
restrictions about the data-type.