Linguistic Components in (Office) Information Systems and a General Evaluation Strategy for Automatic Indexing

Abstract

Abstract The sense or nonsense of improving information systems by linguistic algorithms (mainly by integrating linguistic analysis in the automatic indexing component and by natural language interfaces) have been discussed again and again, yet without an approximation in the viewpoints. It is argued that one reason for this unsatisfactory situation is in the methodic approach. Therefore an alternative complex approach is discussed which was the base of the project EVAL. EVAL evaluates some automatic indexing-systems and a natural language interface. First results of EVAL confirm the methodic approach taken. They are summarized at the end of the paper, which deals mainly with the theoretical background and the methodological approach necessary to answer the question, whether or not linguistic components are profitable for the improvement of existing information systems.