Kernel methods have recently been introduced to solve Natural Language Processing and Text Mining problems. Kernels define a generalised similarity measure between objects of arbitrary structure, with three interesting properties, namely the ability to incorporate prior knowledge about the problem, the implicit mapping of the data into a new feature space, which allows for very richer representation and where problem solving is easier, and finally the independence of learning algorithms from the dimension of this new feature space (—the Kernel trick“). These properties, coupled with robust learning algorithms (for classification, clustering, dimension reduction, filtering, ...) provide some remarkable results in Text Mining tasks, such as document categorization, concept clustering, word sense disambiguation, information extraction, relationship extraction and automatic multilingual lexicon extraction.