Kernel methods for exploratory pattern analysis: A demonstration on text data

Kernel Methods are a class of algorithms for pattern analysis with a number of convenient features. They can deal in a uniform way with a multitude of data types and can be used to detect many types of relations in data. Importantly for applications, they have a modular structure, in that any kernel function can be used with any kernel-based algorithm. This means that customized solutions can be easily developed from a standard library of kernels and algorithms. This paper demonstrates a case study in which many algorithms and kernels are mixed and matched, for a cross-language text analysis task. All the software is available online.