MONO91 Indexes Generated by Machine-Automatic Derivative Indexing chapter Mary Elizabeth Stevens National Bureau of Standards cssenti4ly the original Luhn format, and it should be noted in this connection that while Luhn recognized that the origin of the KWIC principle lay in the making of concordances, he claimed in particular the use of machines to achieve speed, completeness, and accu- racy, and a novel format. 1/ The most common variant to the center position for the indexing window (or keyword position) is at the left or the beginning of the line. Netherwood's selected bibliography of logical machine design, which is probably the first of the modern permuted title indexes to appear in the open literature, used the left-most positions for the index entry word in each title listing. Slant marks were also printed to show the breaks in the normal order of the title (Netherwood, 1958 L437]) A proposed subscription service, advertized in 1958 but never actually brought into operation, would also have used the left-hand position. In these left position examples, the keyword-in-context principle is kept only partially intact since the word in the index position is directly adjacent to its most specific right-hand context, not to its left-hand. In variations such as developed at Stanford Research Institute, however, the index word is extracted from its context and printed separately in the left-hand margin, with the title in its normal order printed to the right. This type of variation has been called "KWOC", for keyword-out-of-context, and is illustrated in Figure 6, which shows the format developed by C. E.I.R., Inc. for the OTS index to U.S. Government Research Reports. Table 1 lists a number of KWIC index projects for which computer programs are or might be made available to interested additional users. Computer programs have been written specifically for the IBM 650, 704, 1620, 709, 7090, and 7094 data processing systems, the G.E. 225 computer, the Deuce Computer in England, the UNIVAC 1103 and 1107 systems, and the Japanese computer JEIPAC, among others. In addition, some permuted title indexes are produced manually, or with the use of simple business office machine equipment. For example, an index to the MBS Bulletin for 1951-1961 has been so produced by the American Institute of Biological Sciences. 3/ 1/ Private communication, excerpt of letter from H.P. Luhn to C. L. Bernier, December 27, 1960: "With respect to the origin of the KWIC Index, you are, of course, right that it is a form of concordance, as stated in my original paper. Furthermore, keyword indexing has been practiced in various forms as far back as a hundred years ago. All of these methods were, however, de- pendent on manual effort. I would say that the significance of the present KWIC Index is based on the fact that it is produced automatically by machine, affording speed of compilation, accuracy and completeness. As far as the particular format of the Index is concerned, this is novel to my knowledge, in accordance with in- formation I have been able to ascertain from others." 2/ "PILOT--a permutation index to this month's literature", see p.8 and Figure 1. A left-most window full-title format was developed at Stanford University in co- operation with the IBM San Jose Laboratories. It has been applied by the Com- putation Center to the titles of computer pro[OCRerr]rams for the benefit of users of the Program Library Computation Center, Stanford University, "The KWIC Index", 1963. See also Marckworth, 1961 L393] National Science Foundation's CR&D Report No. 11, L430], p.10; Janaske, 1962 [299[OCRerr]; Shilling, 1963 Lsso] and [55[OCRerr]J 48 3/