|文件类型：||文章, 计算机文档, 互联网资源|
Carol Jean Godby; OCLC. Office of Research.
|注意：||Title from title screen (viewed Mar. 26, 2004).|
|详述：||Mode of access: Internet.|
|责任：||project manager, Jean Godby.|
This study compares two methods for automatically discovering content-bearing phrases that would improve searchable and browsable indexes of full-text documents. The first method parses the text for noun phrases and uses heuristics to identify the best content-bearing phrases. The second method identifies these phrases by applying frequency and association statistics to the text. It is argued that the second method promises to give results that are just as good as the first, and has the advantages of being conceptually simpler and more generic.