Abstract
A domain-based approach to information retrieval and extraction is presented. The goals of the approach and its relation to other methods are investigated by a unique analysis of the retrieval task. In particular it is argued that the system structure should be highly evocative to users. The method is also complete up to domain equality, over heterogeneous databases. The method allows for direct semi-automatic acquisition of knowledge from the creators of document sources by simple means. The method was realized as a rule-based system and was tested with regard to both retrieval and extraction. Both tasks revealed high levels of precision and recall using simple knowledge-based and domain oriented techniques.