CRANV2 Main test results chapter Cyril Cleverdon Michael Keen Cranfield An investigation supported by a grant to Aslib by the National Science Foundation. Use, reproduction, or publication, in whole or in part, is permitted for any purpose of the United States Government. 78 - CHAPTER 4 Main Test Results Efficiency is something which cannot be achieved without effort expended in the appropriate ways, and measurement is one of the ways towards improved efficiency. It is difficult to see how efficiency can be improved without some basis in measurement. L,T. Wilkins: Social Deviance, page 8. In the two previous chapters have been outlined the environment in which the tests were carried out and the measures which are used for presenting the results. For those who did not wish to work through these two chapters, a brief summary is now given to assist in the interpretation of the test results presented in this chapter. The simplest method of doing this is to illustrate the various points with an example of the tables which present the test results, as in Figure 4.001, to which reference should be made. (1) There are four main groups of index languages and these are: identified by roman numerals: I Single term index languages (eight languages tested) II Simple concept index languages (fifteen languages tested) III Controlled term index languages (six languages tested) IV Abstract and title searches (four languages tested) With each index language there are a number of different recall devices. These are identified by arabic numerals, and the full range of these index languages is set out in Figures 2.5, 2.6 and 2.7. The lower case letter identifies the precision device which is being used. The basic device is coordination, shown by a. The other precision devices are fully explained in Figure 2.8. (2) At the indexing stage, each term was given a rating in relation to its importance in each particular document, and this permits tests to be made with aifferent levels of exhaustiv[ty of indexing, This is to say that searches can be made on the full indexing, which averaged 31 terms for each document, and this is shown as Exhaustivity 3. Alternatively it can be done on a restricted set of terms, where the average was 25 terms per document; this is shown as Exhaustivity 2. Finally searches can be made on the least exhaustive indexing, where the average was 13 terms per document, and this is shown as Exhaustivity 1. (3) Various search rules that were used are explained in Figure 2.9. The basic search rule permitted the combination of any terms and is shown by A. Other search rules are explained in the appropriate section of the tables.