A paradigm for a retrieval effectiveness experiment

Based upon the expectation that empirical research will eventually replace logical attempts to assess the effectiveness of different indexing rationales, the author presents a paradigm for such experimentation. The recommended criterion of retrieval effectiveness is a series of independent measures stated in terms of the user's requirements. The experimental procedures thought necessary to carry out such an experiment are illustrated in the paradigm. In addition, a series of appendices, each of which deals with a specific question or difficulty encountered in this kind of experimentation, is presented.