IRS13 An Analysis of the Documentation Requests chapter E. M. Keen Harvard University Gerard Salton Use, reproduction, or publication, in whole or in part, is permitted for any purpose of the United States Government. x-38 thesaurus, then a sjn[OCRerr]le automatic choice could be made by the system. Using the Abstract Stem versus Abstract Thesaurus result, Figure 22 shows that more of the factors of generality, length or concept frequency are correlated with either dictionary, and even the requests made up by the two preparers do not markedly prefer particular dictionaries. Other criteria may be discovered to aid such a pre-aearch eboice, [OCRerr]d if a perfect choice were achieved the result would be as given in Figure 23, where the curve based on choice of the best dictionary is seen to be better than use of either dictionary exclusively. The possibility of achieving a satisfactory automatic subject re- cognition is considered in an extensive analysis performed by J. O'Connor f7]. It seems certain that some loss of performance due to inability to correctly recognize and match with ideas asked for in requests will be experienced unless very Sophisticated procedures can be developed. However, failure in matching occurs also in manual systems due both to errors and inability to cope with the tasks of manual indexing and vocabulary control; it is thus by no means certain that automatic systems will in practi[OCRerr]e prove inferior.