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. - 14 - In search Type A, any combination of terms was always accepted, without regard for the cases where some combinations accepted might be meaningles_s. For example, consider a question with the search terms Methods, Testing, Analysis, Investigating, Static, Dynamic, Stability, Characteristics, Re-entry, Body, Free, Flight, Tests. At, say,. a coordination level of four, any combination of these search terms would be accepted, such as Methods, Static, I[OCRerr]e-entry, Free. This is only one of many non-sensical combinations of the search terms at this level of coordination. The use of this search rule for investigating nearly every other variable was adopted, since it could be applied with equal consistency to all the different languages, with the exception of the tests of the precision devices of partitioning and inter-fixing on the single ten[OCRerr]n languages. For these tests it was felt that a certain amount of intellect should be put into t.he search rules, and this consisted of a rule (Type B) which did not permit ,subordinate' terms to be accepted unless the associated ,basic' terms was present. The distinction between basic and subordinate terms became apparent when. the single terms of the search questions were grouped into concepts, prior to the test of inter- fixing. In the example mentioned there are certain concepts that would emerge, such as Static stability characteristics, Re-entry body, Methods of testing. Basic terms in these concepts might be Stability, Body and Testing, for these terms are meaningful on their own in the context of the search question. Therefore Search Rule B would require, for instance, that He-entry would not be accepted unless Body was also present, nor would Static be accepted unless Stability was present. The importance of adopting this rule before making a test of interfixing is that at least two terms from a concept must be present for interfixing to be tested. If, in the indexing of a document, the two single and separate terms Static and Stability appeared, and the demand for inter- fixing was added, if they were not interfLxed then only one of the single terms could be accepted (Which would have to be Stability to accord w[OCRerr]th Search Rule B). Without Search Rule B the single term Static could be accepted in this case, or in a case where Stability did not occur at all. Searches C and D were carried out on the single term index languages, and represented an attempt to discover the effect of including more intelligence in searching. The first stage, Search C, involved making a selection of the original starting terms taken from the search question. This was to eliminate from the search prescriptions certain terms such as Problem, Applied, Variation, Influence, Solution, Comparison, Determination, Effect. etc. This search rule was tested on a set of twenty questions, all of which originally had seven starting terms; the selections made resulted in a range of from two to six of the terms, with the average being 4.1. In using these selec.ted sets of search terms, any combination of these was still accepted, as in Search A. Search D used the selected search terms of search C, and made strict