Training a computer to assign descriptors to documents: experiments in automatic indexing

Abstract

During the past five to ten years, increasing interest has developed in the use of machines as a substitution for human intellectual effort in the indexing or classification of the subject content of documents. In terms of practical applications, the greatest interest to date has been in keyword indexing from the significant words actually occurring in titles, abstracts, or full texts. Pioneering use of computers for this purpose, by Luhn and Baxendale, has been followed by the development of a number of KWIC (keyword-in-context) and similar programs.