KISS, a prototype interface to assist users in making relevance judgments, allows them to view appropriate passages within the natural-language text of retrieved documents. It provides tools to incrementally construct phrases based on term adjacency. Iterative selection of appropriate and representative phrases increases the semantic value of user’s constructions. A visualization tool shows the exhaustivity of the coverage in the document set.
KISS has some clear advantages in the interactive process and the access it provides to the basic elements in the retrieved set. The process facilitates the identification of information at the level of specificity needed in the search. The idea of identifying information rather than documents is tightly related to the interface because it provides direct access to individual passages. KISS can also be used to narrow searches, refine query formulation, increase precision and assist users who are working in relevance judgments or information discovery.