The Reuters collection
Mark Sanderson · White Rose Research Online (University of Leeds, The University of Sheffield, University of York) · 1994
This short paper presents the little known Reuters 22,173 test collection, which is significantly larger than most traditional test collections. In addition, Reuters has none of the recall calculation problems normally associated with some of the larger test collections now available. This paper explains the method (derived from Lewis [Lewis 91]) used to perform retrieval experiments on the Reuters collection. Then, to illustrate the use of Reuters, some simple retrieval experiments are also presented that compare the performance of stemming algorithms.