CRANV2 Test Environment 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. -22- two subsets which had 19 questions with 7 starting terms and 17 questions with 11 starting terms. QUESTIONS 1. Relevance assessments, 4 grades 2. Differing number of starting terms and retrieving terms 3. Differing totals of relevant documents 4. Two sources of questions, 'basic' and ,supplementary' 5. Question sets of different sizes, picked according to different criteria, searched on collections of varying sizes. COLLECTION SIZE 1. 1400 documents 2. 350 documents from the 1400,documents. 3. 200 documents from the 350 document subset. SUBJECT TERMINOLOGY 1. Aerodynamics 2. Aircraft Structures. FIGURE 2.11 SUMMARY OF MAIN ENVIRONMENTAL FACTORS By the time we came to investigate the simple concept languages and the controlled term languages, the clerical effort involved in carrying out searches precluded the use of the full sets of questions, and accordingly a set of 42 questions was prepared, consisting entirely of questions in the field of aerodynamics. It is this set which is used for presenting the majority of the test results in Chapter 4. At a later stage, this subset was extended to 77 questions in the field of aerodynamics; finally an additional set of 42 questions in the field of structures was compiled for purposes of comparison, with the aerodynamic question set of similar sizes. The subsets of questions are all numbered, and details of these appear in Fig. 2.12. Lists of the question numbers for subsets 1, 2 and 3 were given in Vol. I, Appendix 3E; the remaining subsets are shown in Appendix 3.2 of this volume. Reduced collection sizes were also used for reasons of the effort involved in testing. This was not only the clerical effort involved in the searching, but also the intellectual effort involved in compiling word lists for the various index languages. When it was decided to test simple concepts, a set of 200 documents was chosen, and the initial task involved re-formulating the indexed concepts from the original