Salton, GerardZhao, ZhongnanBuckley, Chris2007-04-232007-04-231990-07http://techreports.library.cornell.edu:8081/Dienst/UI/1.0/Display/cul.cs/TR90-1137https://hdl.handle.net/1813/6977A syntactic approach is described for generating indexing phrases usable for the content identification of natural-language texts. The phrase generation method is based on a simple language analysis system that determines the syntactic function of individual text words with a high degree of accuracy, and chooses of indexing phrases based on weights assigned to the phrase components. The proportion of phrases that appear to be acceptable for content identification ranges from 96 to 98 percent.1827851 bytes442318 bytesapplication/pdfapplication/postscripten-UScomputer sciencetechnical reporttechnical report