The amount of electronically stored textual information is continuously increasing both on the internet and in company assets, and there are no good solutions to easily locate the most needed information. Because search engines do not take into account the meaning of the word and its context, in the end the user has to select the right information from the unstructured result set. If the text is annotated and linked to the ontology of the annotation, then the user can directly navigate along the links of the semantic annotation to the desired information.

In this paper we present a software framework to semi automatically generate a semantic representation of the knowledge of the Networkshop conference series and display on a web portal the generated ontology together with the references to the occurrences of the instances in the source text. The framework presented in this paper makes advances in the following fields: we do not assume that the source text has uniform and formally
defined structure, we address English and Hungarian text as well, we incorporate machine learning techniques in the process, and provide a flexible content management system for the presentation of the generated Topic Map on a web based portal.