Automatic and manual indexing performance in a small file of medical literature

A test of the performance of an automatic indexing procedure based on the processing of the full text of medical journal articles shows results comparable to those achieved using Excerpta Medica and MEDLARS manual indexing. A file of nineteen documents was partitioned by sixteen questions, for each of which two search strategies in each indexing language were formulated. The results were compared to ideal answers chosen by a fourth-year medical student who read the entire file and posed the questions.