Text Mining has established itself as a valuable tool for knowledge extraction in many commercial and scientific areas. Accordingly, a large number of different methods have been developed focusing on a broad range of different tasks. We report on a novel system architecture that is fundamentally service-based, i.e., it models and implements text mining and knowledge extraction routines as independent, yet federated services. The system has several layers: (1) Base services perform various fundamental extraction tasks. They all implement a fixed interface but keep their particular algorithms and functionality. (2) A metaservice acting as a central access point to those base services, thus providing a homogeneous interface to different algorithms. (3) An aggregation service on top of the metaservice which implements functionality to graphically show, compare, and aggregate the results of different base services. Each layer is accessible as a Web Service and thus ready to be integrated in applications that are higher up in the value chain, such as authoring tools or systems for the automatic construction of knowledge bases. We developed our system with a focus on the mining of Life Science text collections. It is available from http://www.bc-viscon.net.