Short cuts.

Text-summarization computer programs are an example of “natural-language processing,” in which computers process information expressed in a real human language rather than a computer language. Summarizing involves the two related tasks of making sense of words presented to the computer and creating...

Descripción completa

Detalles Bibliográficos
Publicado en:Economist Vol. 333; pp. 85 - 87
Formato: Artículo
Publicado: Economist Newspaper Limited 12/17/1994
Acceso en línea:Ver este registro en EBSCOhost
Descripción
Sumario:Text-summarization computer programs are an example of “natural-language processing,” in which computers process information expressed in a real human language rather than a computer language. Summarizing involves the two related tasks of making sense of words presented to the computer and creating fresh text that also makes sense. A statistical approach is particularly suited to dealing with natural-language input. Statistical tools search the text to locate phrases that occur most often. The program takes these phrases as particularly important and employs them as the basis of subsequent work. An example is provided of a BT program that identifies the important phrases in a piece, then reexamines the whole piece, assigning each sentence a score based on the appearance within it of differently weighted phrases; and summarizes by selecting the highest-scoring sentences, using phrases that it already has to hand in order to produce a natural-language output that reflects its input.