| 摘要 |
PROBLEM TO BE SOLVED: To precisely specify keywords characterizing each document, and to grasp the contents of each document at a glace in a document database in which a plurality of documents related with a specific field are summarized. SOLUTION: This keyword extraction method in a document database is provided for making a programmed computer execute a step for acquiring the whole number m of terms included in a document database in which n pieces of documents related with a specific field are summarized and the respective terms T<SB>j</SB>(<SB>j</SB>=1, 2, 3, ..., m), and for managing the identification of the respective terms T<SB>j</SB>, a step for calculating appearance frequency W<SB>ij</SB>related with the terms T<SB>i</SB>in a document D<SB>i</SB>by a predetermined calculation formula, a step for calculating distribution S<SP>2</SP><SB>j</SB>of the appearance frequency W<SB>ij</SB>value concerning the terms T<SB>j</SB>, a step for calculating significance V<SB>ij</SB>of the terms T<SB>j</SB>in the document D<SB>i</SB>by V<SB>ij</SB>=U<SB>ij</SB>×S<SP>2</SP><SB>j</SB>by using the appearance frequency of the terms T<SB>j</SB>in the document D<SB>i</SB>as U<SB>ij</SB>and a step for preparing and outputting a term list in which the terms T<SB>j</SB>are listed up based on the V<SB>ij</SB>. COPYRIGHT: (C)2006,JPO&NCIPI
|
