Today's Web may be considered to consist of a largely unstructured collection of documents. The Semantic Web is an effort to give information a well defined meaning so that it may be interpreted by computers. On the Web this means that metadata may be published in documents so that they may be processed in a variety of useful ways by machines. An important type of such metadata is classification information, the process of classifying documents according to some taxonomy which facilitates useful functionality on the Web such as searching and browsing. This thesis investigates machine learning approaches to classifying text within the framework of the Semantic Web. It endeavours to advance Semantic Web research by providing a working application of Semantic Web technology thereby pinpointing limitations in the current framework and suggesting where it might be improved. This work also sets out the case for text classifiers within the Semantic Web; the requirements for this to be a working technology and, in particular, it introduces the case for Web service generation by generating classifier services on the fly using arbitrary tagged corpora.