Research and evaluation in information retrieval

This paper is a discussion document, drawing on the experiences of the Okapi team as reported elsewhere in this issue. It is intended to raise some of the issues that are currently exercising the information retrieval research community, in the context of experimentation and evaluation.