CRANV1P1 Formation of Index Languages chapter Cyril Cleverdon Jack Mills 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. - 88 - 'Gas welding &' and 'Pressure welding &' (the ampersands mean that the two broader terms Gas welding and Pressure welding can be jointly substituted for the single and more precise term Pressure gas welding). This procedure led to a number of simi- lar new references; e.g., Prandtl-Meyer flow Use Supersonic flow x Expansion. However, this device leads to a difficulty, inherent in post coordinate indexing, when these particular coordinations generate new reference structures of their own which are not apparent at the level of single isolated terms. E.J.C. offers no guidance on this point. For example, assume that for Conical flow the instruction is to use Cones x Flow. In this case, some sort of reference seems desirable, either from the rejected phrase Conical flow or from its constituent terms in the form: Cones: when coordinated with Flow, RT Shock waves. In this particular case an intermediate connective was established by using a heading Mach cones. Examples of where the reference structure could be fairly elaborate are Flow x Parameters and Vapour x Screens x Procedures. Each of these subjects has its own set of related terms, generated entirely by the conjunction of their con- stituent terms, e.g. Vapour screen method, which now has its own related terms, such as Carbon tetrachloride vapour, Humidity control, Temperature control, Oper- ational fog density. The implied need for such connective references if the syndetic structure were to be developed raised problems of complexity in the scan-column search techniques (which were designed to be purely clerical in operation). In view of this, together with the fact that E. J. C. quite ignored such post coordinate reference needs, it was decided to follow the E. J.C. policy and rely on future analysis of searches to show where this weakness contributed to failures in performance. E.J.C. roles The E. J.C. system of roles Which appear, without explanation, in Table 3 of Ref.28was used in the indexing. The reasons why roles were not tested in earlier languages have already been discussed, but the availability of a ready-made set of roles seemed a useful opportunity to investigate whether some of the assumptions made there (those relating to the applicability of roles to aerodynamic literature) were justified. Unfortunately, at this stage of the project, the time factor was begin- ning to limit the amount of new testing which could be undertaken and it was decided that this validation was not possible. However the decision was not made umil the tentative examination of the feasibility of using the roles had been undertaken by adding them to the indexing descriptions of a small sample of reports. These, in fact provided examples of most of the objections and difficulties we had already met in the earlier attempts to use roles. An example of the difficulties inherent in using the roles may be seen if we consider a particular document and some of the questions to which it had been judged relevant. In a docu- ment (1014) on the application of piston theory and the study of aeroelastic problems, one of the themes indexed was calculation of panel flutter in supersonic flow by piston theory. Some of the problems immediately raised by the addition of roles, taking particular terms, were as follows:- Panel is clearly a patient (role 9), but could conceivably be regarded also as a sup- port or host in a process (role 5). Vibration (the E.J.C. term for flutter) is un- - doubtedly an undesirable component (role 3) in the aerodynamic behaviour of the panel.