The emerging conception of a “digital library” as an information environment in which distributed information sources are made available in an integrated way implies new requirements for indexing and retrieval methods. Therefore, one important research topic in the field of digital libraries is the design of task-specific methods for very precise searching. In addition, we have to anticipate highly dynamic information bases which change continually. This paper outlines an approach which addresses both problems. Rule-based indexing can be used to support highly precise searches in dynamic information bases. The method enables users to perform conceptual searches in image collections, thus going far beyond contemporary content-based retrieval methods for pictorial material. The paper concludes with a discussion of the empirical results obtained so far.