The identification and analysis of the knowledge available in document form is a key element of corporate knowledge management. In engineering-intensive organizations, it involves tasks such as standard generation and evaluation, comparison of related cases and experience reuse in their treatment. No single information retrieval technique is likely to adequately deal with such tasks independent of the specific situation, and no fully automatic approach is likely to yield good results. In this paper, we therefore present the design, implementation, and some application experiences with a modular approach that allows a variety of techniques from semantic document analysis to interoperate with a tailorable map-centered visualization of the structure of technical document collections.