The Significance Of The Cranfield Test On Index Language By Mitsie W Cleverdon Analysis

Cyril W. Cleverdon’s 1991 article, “The Significance of the Cranfield Tests on Index Languages,” outlines the series of experimental tests into information retrieval the author took part in through the 1950s and 1960s while working as a librarian at the Cranfield College of Aeronautics in England. In the post-war midst of lifted security restrictions on many scientific and technical reports, the need for an accurate and efficient method of indexing was acute, and many new techniques were idealized in the ‘50s. Noting the lack of firm data supporting or opposing these methods, Celverdon set out to systematically compare them and their parts, in laboratory-based tests that became known as Cranfield 1 and Cranfield 2. Cranfield 1 explored four
7). The documents’ authors were asked to provide questions, topics, and relevance judgements based on their papers, which were used along with thesauri to create a database with which to compare retrieval rates. With Cranfield 2, Cleverdon introduced the concepts of recall and precision, both dependent on relevance. By manipulating the many decisions that go into the indexing process, Cleverdon was able to study the effects of specificity and exhaustivity. Having the same inverse relationship as recall and precision, all were important to striking the delicate balance the led to optimum retrieval
Known as the Cranfield paradigm, this scientific approach assumes that “information systems may be analysed in isolation from the real messy world of users, information needs, and so on; and moreover that the parts of systems can also be analyzed independently” (Bawden & Robinson, 2012, p. 43). By conducting tests under highly controlled conditions, Cleverdon was able to quantitatively analyse results, thereby providing a significant scientific contribution to the field of information studies at the time. Bawden and Robinson further recognise his studies as “the first time in which information organization and retrieval tools had been analysed and tested empirically, rather than being discussed in a theoretical and philosophical way” (p.