MONO91 Other Potentially Related Research chapter Mary Elizabeth Stevens National Bureau of Standards Another partial approach to applying syntactic analysis techniques to automatic indexing is based upon syntactic word-class recognitions. Giuliano and his associates at Arthur D. Little, Inc. , (1963 [230]), have investigated on a small-scale basis the use of the Kuno-Qettinger programs developed at Harvard for this purpose (Kuno and Oettinger3 1963 [340]). The broad program of information and language data processing research at System Development Corporation specifically includes investigations of structural patterns of sentences at the syntactic level and also of semantic factors such as the studies of polysemy and homographic ambiguity by Doyle, Wasser, and others. 3orko reports: We. are analyzing actual written text for multiple meanings. . . The data for this study were drawn from the corpus of 618 psychological abstracts. Tabulations of frequency of paired and single word listings were used. A number of corpus-derived word frames have been prepared. Although this research is still in its early phase, we feel that we have made a good start on the problems of semantic analysis." 1/ In Czechoslovakia, at the Karlova Universita, both statistical and semantical methods for' automatic abstracting are reported as being under consideration.2/ Other examples of proposals for the use of syntactic analysis techniques for the improvement of automatic indexing products include those of Spangler, Levery, Plath, Thorne, and Climenson and his colleagues at RCA, as well as the suggestions of those whose interests in automatic syntactic analysis have been primarily directed to problems of machine translation or more general problems of linguistic analysis. Hays, for example, although principally concerned with MT, indicates that the methods for determining phrase structures have obvious applications to the automatic determination of categories useful in the indexing of documents. 3/ An existing GE-225 computer program for KWIC-typ e indexing from both titles and abstracts at General Electric's Phoenix Laboratories is being extended to incorporate word analysis features taking into account both syntactic and semantic aspects of a given line or sentence of text. 4/ Levery provides an example of similar directions being ex- plored in European research, more generally oriented toward linguistic considerations as such than to machine-derivable criteria (largely statistical to date), which seek to combine the benefits of both human and machine processes by way of automatic syntactic analyses. He claims) for example, that: 1/ 2/ 3/ 4/ Borko, 1962 [75], p. 6. National Science Foundation's CR&D report, No. 11 [430], p. 123. See Hays, 1961, [258], p. 13: ". . Two broad problems on which work is just beginning at RAND: grammatic transformations and distributional semantics. The latter problems are especially important for automatic indexing, abstracting, and text searching." See also de Grolier, 1962 [152], p. 137. NationalScience Foundation'SCR&D report No.11 [430], p.21. 129